[PATCH v4 6/8] ash: use BB_EXECVPE to execute commands with FEATURE_SH_STANDALONE

[Nadav Tasher]

On Wed, Jan 29, 2025 at 09:14:59AM +0000, Ron Yorston wrote:
> Nadav Tasher <tashernadav at gmail.com> wrote:
> > static void
> >-tryexec(IF_FEATURE_SH_STANDALONE(int applet_no,) const char *cmd, char **argv, char **envp)
> >+tryexec(const char *cmd, char **argv, char **envp)
> > {
> > #if ENABLE_FEATURE_SH_STANDALONE
> >-	if (applet_no >= 0) {
> >-		if (APPLET_IS_NOEXEC(applet_no)) {
> >-			clearenv();
> >-			while (*envp)
> >-				putenv(*envp++);
> >-			popredir(/*drop:*/ 1);
> >-			run_noexec_applet_and_exit(applet_no, cmd, argv);
> >-		}
> >-		/* re-exec ourselves with the new arguments */
> >-		execve(bb_busybox_exec_path, argv, envp);
> >-		/* If they called chroot or otherwise made the binary no longer
> >-		 * executable, fall through */
> >-	}
> 
> This removes the call to popredir() without providing any replacement.
> Since it's an internal ash function it can't be called from BB_EXECVPE.
> 
> The call was introduced in commit b31b61bb9b with this description:
> 
>   ash: fix redir_leak.tests if STANDALONE=y
> 
>   If STANDALONE and we run a NOEXEC applet, saved copies of redirected fds
>   were visible for the child. They have CLOEXEC bit, yes, but we do not exec
>   in this case.
> 
> Although the patch set doesn't touch hush, the _other_ shell in BusyBox,
> it has a similar requirement.  Before running a NOEXEC applet it calls
> two internal functions:  close_saved_fds_and_FILE_fds() and
> switch_off_special_sigs().
> 
> Cheers,
> 
> Ron
Hi!

The first issue was addressed in the 5th version of the patchset.

Regarding hush, do you think I should also implement this feature there?

Nadav