[Bug libc/28853] Problems with terminal control

["carlos at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=28853

Carlos O'Donell <carlos at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |carlos at redhat dot com

--- Comment #4 from Carlos O'Donell <carlos at redhat dot com> ---
Dave,

Thanks for the bug report.

I can confirm that posix/tst-spawn6 breaks out of the support/* isolation and
in doing so impacts the shell / process group / session on the terminal.

e.g.
[2]+  Stopped                 less ~/build/glibc-review/check.log

(In reply to John David Anglin from comment #2)
> I suppose this is the problematic part of the change:
> 
>   /* Set the controlling terminal.  */
>   if ((attr->__flags & POSIX_SPAWN_TCSETPGROUP) != 0)
>     {
>       /* Check if it is possible to avoid an extra syscall.  */
>       pid_t pgrp = (attr->__flags & POSIX_SPAWN_SETPGROUP) != 0
>                     && attr->__pgrp != 0
>                    ? attr->__pgrp : __getpgid (0);
>       if (__tcsetpgrp (attr->__ctty_fd, pgrp) != 0)
>         goto fail;
>     }

This is not problematic per-se, since this is what you want to do for shell
control, but if the test isolation code doesn't stop this affecting your
current shell then that's a problem.

The posix/tst-spawn6 test is the *only* test in which we exercise process group
control testing, and so it may break the testing isolation we expect.

(In reply to John David Anglin from comment #3)
> The  function tcsetpgrp() makes the process group with process group ID
> pgrp the foreground process group on the  terminal  associated  to fd,
> which  must  be  the  controlling  terminal of the calling process, and
> still be associated  with  its  session.   Moreover,  pgrp  must  be a
> (nonempty)  process  group belonging to the same session as the calling
> process.
> If tcsetpgrp() is called by a member of a background process  group in
> its  session, and the calling process is not blocking or ignoring SIGT-
> TOU, a SIGTTOU signal is sent to all members of this background process
> group.

Yes, for the current /dev/tty (in the test), we make the posix_spawn'd
process the foreground process group for the session to verify the test works
correctly.

I can't easily containerize the test because we don't setup an alternative
TTY/session/pgrp in the container e.g. tests-container += tst-spawn6.

make[2]: Leaving directory '/home/carlos/src/glibc-review/posix'
FAIL: posix/tst-spawn6
original exit status 1
error: tst-spawn6.c:138: open64 ("/dev/tty", 0x0, 0600): No such file or
directory
error: 1 test failures
make[1]: Leaving directory '/home/carlos/src/glibc-review'

-- 
You are receiving this mail because:
You are on the CC list for the bug.