Security bugs are reported here via bugzilla

[bugzilla-daemon AT metalab.unc.edu]

http://bugs.sourcemage.org/show_bug.cgi?id=4272





------- Additional Comments From seth AT tautology.org  2003-09-08 00:45 -------
I'm currently working on init.d defaulting to no.  I have it almost all done, 
in
fact, just need to make some sort of temporary state cache to know the old
enabled status of the old init script before the script gets dispelled.

Check this code that's already in libinit.d, the comment on -o true is in 
there
because I was going to have dispel not dispel on certain conditions, but I 
think
the idea of dropping something into /tmp or /var/state/sorcery before a
cast-triggered dispel (not a real dispel-triggered dispel) that has the old
enabled value of the init script.  For xinetd, I can do something similar, but
yeah, I totally agree with this bug (and have been working to fix it).  I
wouldn't need any "ESSENTIAL" logic with xinetd because, really, xinetd isn't
needed for getting a basesystem working.

==>
      if ! grep -Fq 'ESSENTIAL=yes' $script_path ; then
        if [ -x $runlevel_dir/$script_name -o true ] ; then # -o true is in
there until dispel doesn't delete the old init scripts
          query "Shall the $script_name init script be enabled on install?" y 
&&
ENABLE_SCRIPT=y || ENABLE_SCRIPT=n
        else
          query "Shall the $script_name init script be enabled on install?" n 
&&
ENABLE_SCRIPT=y || ENABLE_SCRIPT=n
        fi
<==

I'm thinking /var/state/sorcery/init.d.$spell.enabled could be created and 
have
it contain 1 if it was enabled before, nothing if it wasn't, and do that 
during
dispel for both init.d and xinetd so we can do rebuilds properly.  If somebody
can beat me to it, they are more than welcome (or offer alternative
implementation ideas).  I'll try and get this done tomorrow, but if I don't, 
or
if anybody else wants to do it, any volunteers?



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.