[4711] in Release_7.7_team

home help back first fref pref prev next nref lref last post

Another Red Hat bogon (in bash)

daemon@ATHENA.MIT.EDU (Greg Hudson)
Sat Jul 24 13:38:26 2004

Date: Sat, 24 Jul 2004 13:38:21 -0400
Message-Id: <200407241738.i6OHcLSA006864@egyptian-gods.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@MIT.EDU

If you refer to:

  https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=86514

you'll see that someone reported a bash bug related to a piece of
apparently bogus bash code:

  static int
  redir_open (filename, flags, mode, ri)
  [...]
        fd = open (filename, flags, mode);
  #if defined (AFS)
        if ((fd < 0) && (errno == EACCES))
          fd = open (filename, flags & ~O_CREAT, mode);
  #endif /* AFS */

(bash has no version control information that I'm aware of, so I can't
figure out why this code would be appropriate.)

Red Hat's response was to stop building bash with AFS support.
Unfortunately, that disables some other, more useful code, such as
code to make "[ -r" use access() instead of stat bits.  As a result,
some locker software (e.g. firefox) breaks under 9.3 because it uses
/bin/sh scripts which use "[ -r" code.

We're going to have to live with this for 9.3, but it may continue to
dog us as people install new locker software with an 077 umask and use
/bin/sh scripts to poke around in the locker.

What's the best way to put pressure on them over this?  I can leave a
comment in #86514 noting that some of their paying customers use AFS
and that they should back out their change and use a more tailored fix
(i.e. patching bash to remove the offending redir_open code).

home help back first fref pref prev next nref lref last post