Re: [PATCH] Fix getlogin() to check only stdin to get a valid tty

[Corinna Vinschen <vinschen@redhat.com>]

On Jul 13 15:01, Jeff Johnston wrote:
> I think that when in doubt, going with glibc is the right answer.
> 
> I'll give Corinna time to pipe in if she wishes, otherwise, I'll check in
> the glibc-like patch.

Sounds right to me.


Corinna

> 
> -- Jeff J.
> 
> On Thu, Jul 13, 2023 at 12:58 PM Jordi Sanfeliu via Newlib <
> newlib@sourceware.org> wrote:
> 
> > Hello,
> >
> > So we have two possible patches to solve this situation:
> >
> > My first original patch that only checks if stdin is a tty and seems to
> > match with glibc comments provided by Torbjörn. The only problem is that
> > it won't work in the scenario when someone pipes some data on stdin to
> > the application:
> >
> > $ echo | ./test
> > $
> >
> >
> > diff --git a/newlib/libc/unix/getlogin.c b/newlib/libc/unix/getlogin.c
> > index da4f47a95..e646bcb08 100644
> > --- a/newlib/libc/unix/getlogin.c
> > +++ b/newlib/libc/unix/getlogin.c
> > @@ -16,9 +16,7 @@ getlogin ()
> >     extern char *ttyname ();
> >     char *tty;
> >
> > -  if (((tty = ttyname (0)) == 0)
> > -      || ((tty = ttyname (1)) == 0)
> > -      || ((tty = ttyname (2)) == 0))
> > +  if ((tty = ttyname (0)) == 0)
> >       return 0;
> >
> >     if ((utmp_fd = open (UTMP_FILE, O_RDONLY)) == -1)
> >
> >
> >
> > And a patch based on the comments of Torbjörn that checks all three fds
> > before returning NULL:
> >
> > diff --git a/newlib/libc/unix/getlogin.c b/newlib/libc/unix/getlogin.c
> > index da4f47a95..5a3f172e7 100644
> > --- a/newlib/libc/unix/getlogin.c
> > +++ b/newlib/libc/unix/getlogin.c
> > @@ -16,10 +16,10 @@ getlogin ()
> >     extern char *ttyname ();
> >     char *tty;
> >
> > -  if (((tty = ttyname (0)) == 0)
> > -      || ((tty = ttyname (1)) == 0)
> > -      || ((tty = ttyname (2)) == 0))
> > -    return 0;
> > +  if ((tty = ttyname (0)) == 0)
> > +      if ((tty = ttyname (1)) == 0)
> > +          if ((tty = ttyname (2)) == 0)
> > +              return 0;
> >
> >     if ((utmp_fd = open (UTMP_FILE, O_RDONLY)) == -1)
> >       return 0;
> >
> >
> > Any thoughts?
> > Thanks.
> >
> >
> > On 7/13/23 18:25, Torbjorn SVENSSON wrote:
> >
> > > I took a sneak peak at how glibc does this and there is this comment:
> > >
> > >    /* Get name of tty connected to fd 0.  Return NULL if not a tty or
> > >       if fd 0 isn't open.  Note that a lot of documentation says that
> > >       getlogin() is based on the controlling terminal---what they
> > >       really mean is "the terminal connected to standard input".  The
> > >       getlogin() implementation of DEC Unix, SunOS, Solaris, HP-UX all
> > >       return NULL if fd 0 has been closed, so this is the compatible
> > >       thing to do.  Note that ttyname(open("/dev/tty")) on those
> > >       systems returns /dev/tty, so that is not a possible solution for
> > >       getlogin().  */
> > >
> > > Based on this comment, I guess it would be sane to drop the check on
> > > stdout and stderr, but it would have the consequence that you are not
> > > able to pipe some data on stdin to the application that calls getlogin
> > > as it would fail in that scenario.
> > >
> > > I'm not a maintainer of newlib so I don't really have anything to say
> > > about what path you decide to go.
> >
> > --
> > Jordi Sanfeliu
> > FIBRANET Network Services Provider
> > https://www.fibranet.cat
> >
> >