Re: patched garball for gdm



That is a reasonable workaround.  However, that means that 

	Authentication scheme: verify-shadow

is broken on Debian-based systems.  

As reported by both Briggs and Drach, there are problems with undefined
references in

        ./daemon/verify-shadow.c

Specifically:

        * ve_string_empty
        * GdmAllowRoot, GdmAllowRemoteRoot, GdmDisplayLastLogin,
        GdmRetryDelay


On systems where PAM is chosen as the authentication scheme 

        ./daemon/verify-shadow.c

is never compiled so the problem never arises. 

ve_string_empty is a macro defined in

        ./vicious-extensions/ve-misc.h

adding this include file to ./daemon/verify-shadow.c

        #include "gdm.h"
        #include "misc.h"
        #include "slave.h"
        #include "verify.h"
        #include "errorgui.h"
        #include "ve-misc.h"   <<========  

will eliminate the undefined reference to ve_string_empty complaint.


GdmAllowRoot, GdmAllowRemoteRoot, GdmDisplayLastLogin, GdmRetryDelay
were all 

        * declared "extern" in verify-shadow.c 
        
and are 

        * defined as static types in gdmconfig.c. 

The "static" storage class specifier is the problem as it restricts the
scope of Gdm* to gdmconfig.c. The set of static variables will *not* be
exported to the linker.

What to do???

I would edit verify-shadow.c 

and replace

        /* Configuration option variables */
        extern gboolean GdmAllowRoot;
        extern gboolean GdmAllowRemoteRoot;
        extern gint GdmRetryDelay;
        extern gboolean GdmDisplayLastLogin;

with 

        /* Configuration option variables */
        static gboolean GdmAllowRoot = FALSE;
        static gboolean GdmAllowRemoteRoot = FALSE;
        static gint GdmRetryDelay = 0;
        static gboolean GdmDisplayLastLogin = TRUE;

which should (a) work, and (b) do no harm.

I would appreciate hearing if this works.


-Joseph


===============================================================================
On Sat, 2005-11-26 at 08:50 -0600, J. Gardner Biggs wrote:
> On Fri, 2005-11-25 at 22:35 -0500, Joseph E. Sacco, PhD wrote:
> > Carlos Gabriel Drach <carlos drach gmail com> reported similar problems
> > with debian sarge.
> > 
> > I posted an approach yesterday that will eliminate the ve_string empty
> > problem and begin to probe the source of Gdm* problem.
> > 
> > Take a look and see if that helps.
> > 
> > 
> > -Joseph
> 
> 
> that did the trick, I just needed to install the pam development
> packages.
> 
> Thanks.
> > =================================================================================

> 
-- 
joseph_sacco [at] comcast [dot] net




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]