On Tue, Jun 16, 2015 at 03:54:09PM -0600, Jon Trulson wrote:
> On Mon, 15 Jun 2015, Isaac Dunham wrote:
> 
> >While this change is functionally equivalent to the original code
> >(apart from the detail that it's actually valid), I wonder if we
> >should try to update to match current xdm or make this do the "right"
> >thing...well, it looks like xdm eliminated this code over a decade
> >ago.
> >
> >I see that neither of the CVEs for xdm apply for most builds;
> >but for "AlphaArchitecture", where SIA is defined, seteuid(0)
> >is called without error checking...I suppose that cannot be a
> >privilege escalation, but rather might continue despite failing
> >to elevate privileges.
> 
> Ok to apply this patch then?  It looks like particularly crappy code
> to begin with, so I'm ok with removing it.

Yes.

> As for seteuid() - yes that should be checked, but if it fails, it
> fails, so I do not see how that could cause a priv escalation...
> [...]

One of the CVEs for XDM was that "setuid and seteuid were called without
error checking, leading to a potential privilege escalation" 
(summary, not exact).
If one calls seteuid(regular_user), it fails, and one proceeds as if it
had succeeded, code that *should* be running with the permissions of
regular_user may be running as root instead; this is the only way to
get a privilege escalation.
However, that's the reverse of what's happening on Alpha:
seteuid(root) might fail, and then code that was written to run as root
may run with lower permissions.

HTH,
Isaac Dunham

------------------------------------------------------------------------------
_______________________________________________
cdesktopenv-devel mailing list
cdesktopenv-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cdesktopenv-devel

Reply via email to