Re: [Emacs-diffs] master 0c94b84: * nt/inc/ms-w32.h (execve) [MINGW_W64]: Make commentary more accurate.

[Eli Zaretskii <eliz@gnu.org>]

> From: Ken Brown <kbrown@cornell.edu>
> Cc: Angelo Graziosi <angelo.graziosi@alice.it>
> Date: Thu, 1 Sep 2016 15:57:00 -0400
> 
> On 9/1/2016 1:16 PM, Eli Zaretskii wrote:
> > +   However, using the prototype with intptr_t causes GCC to emit
> > +   warnings.  Fortunately, execve is not used in the MinGW build, but
> > +   the code that references it is still compiled.  */
> 
> Wouldn't it be easier to prevent that code from being compiled?

It would for the Windows build, but it would add one (actually more,
see below) #ifdef WINDOWSNT into the mainline code.  AFAIR, Paul
(CC'ed) wanted to keep those to a minimum, so I preferred not to ifdef
away the code.

> I think the following would suffice:
> 
> --- a/src/sysdep.c
> +++ b/src/sysdep.c
> @@ -146,6 +146,7 @@ disable_address_randomization (void)
>   }
>   #endif
> 
> +#ifndef WINDOWSNT
>   /* Execute the program in FILE, with argument vector ARGV and environ
>      ENVP.  Return an error number if unsuccessful.  This is like execve
>      except it reenables ASLR in the executed program if necessary, and
> @@ -170,6 +171,7 @@ emacs_exec_file (char const *file, char *const 
> *argv, char *const *envp)
> 
>     return err;
>   }
> +#endif /* not WINDOWSNT */

That whole function is not used on Windows, so if we are to do this,
the following fragment of main in emacs.c should also be ifdefed away:

    /* True if address randomization interferes with memory allocation.  */
  # ifdef __PPC64__
    bool disable_aslr = true;
  # else
    bool disable_aslr = dumping;
  # endif

    if (disable_aslr && disable_address_randomization ())
      {
	/* Set this so the personality will be reverted before execs
	   after this one.  */
	xputenv ("EMACS_HEAP_EXEC=true");

	/* Address randomization was enabled, but is now disabled.
	   Re-execute Emacs to get a clean slate.  */
	execvp (argv[0], argv);

	/* If the exec fails, warn and then try anyway.  */
	perror (argv[0]);
      }

If Paul doesn't mind (nor anyone else), we can certainly do that.

Thanks.