Please Review: (514824) Multi-macro ACIs can cause double free if macro attribute does not exist

[Nathan Kinder]

https://bugzilla.redhat.com/show_bug.cgi?id=514824

[Nathan Kinder]

On 07/30/2009 03:37 PM, Nathan Kinder wrote:
>  https://bugzilla.redhat.com/show_bug.cgi?id=514824
>  ------------------------------------------------------------------------
>
>  --
>  389-devel mailing list
>  389-devel(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-directory-devel
>      
Here's a new patch that addresses some issues Noriko brought up.

The list "a" was being set to NULL when we found an attribute match for a
macro, but this is no longer necessary now that we reset "a" to NULL 
when the
memory is handed off to the working_list (which covers both the cases of
finding/not finding the attribute).

We were also accessing element 0 of list "a" right after handing the 
memory off
to the working_list, but we weren't checking if "a" was NULL first.  I
don't
believe that "a" could be NULL at this point, but it's safest to check 
first in
case there is some corner case we're not considering.

-NGK

[Noriko Hosoi]

On 07/30/2009 05:10 PM, Nathan Kinder wrote:
>  On 07/30/2009 03:37 PM, Nathan Kinder wrote:
> > https://bugzilla.redhat.com/show_bug.cgi?id=514824
> > ------------------------------------------------------------------------
> >
> > --
> > 389-devel mailing list
> > 389-devel(a)redhat.com
> > https://www.redhat.com/mailman/listinfo/fedora-directory-devel
> >    
>  Here's a new patch that addresses some issues Noriko brought up.
>
>  The list "a" was being set to NULL when we found an attribute match for a
>  macro, but this is no longer necessary now that we reset "a" to NULL 
>  when the
>  memory is handed off to the working_list (which covers both the cases of
>  finding/not finding the attribute).
>
>  We were also accessing element 0 of list "a" right after handing the 
>  memory off
>  to the working_list, but we weren't checking if "a" was NULL first.  I 
>  don't
>  believe that "a" could be NULL at this point, but it's safest to check 
>  first in
>  case there is some corner case we're not considering.
>
>  -NGK
>  ------------------------------------------------------------------------
>
>  --
>  389-devel mailing list
>  389-devel(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-directory-devel
>      
Looks good.  Thanks for considering my questions!
--noriko

[Nathan Kinder]

On 07/30/2009 05:15 PM, Noriko Hosoi wrote:
>  On 07/30/2009 05:10 PM, Nathan Kinder wrote:
> > On 07/30/2009 03:37 PM, Nathan Kinder wrote:
> >> https://bugzilla.redhat.com/show_bug.cgi?id=514824
> >> ------------------------------------------------------------------------
> >>
> >> --
> >> 389-devel mailing list
> >> 389-devel(a)redhat.com
> >> https://www.redhat.com/mailman/listinfo/fedora-directory-devel
> >>    
> > Here's a new patch that addresses some issues Noriko brought up.
> >
> > The list "a" was being set to NULL when we found an attribute match for a
> > macro, but this is no longer necessary now that we reset "a" to NULL 
> > when the
> > memory is handed off to the working_list (which covers both the cases of
> > finding/not finding the attribute).
> >
> > We were also accessing element 0 of list "a" right after handing the 
> > memory off
> > to the working_list, but we weren't checking if "a" was NULL first.  
> > I don't
> > believe that "a" could be NULL at this point, but it's safest to 
> > check first in
> > case there is some corner case we're not considering.
> >
> > -NGK
> > ------------------------------------------------------------------------
> >
> > --
> > 389-devel mailing list
> > 389-devel(a)redhat.com
> > https://www.redhat.com/mailman/listinfo/fedora-directory-devel
> >    
>  Looks good.  Thanks for considering my questions! 
Pushed to master.  Thanks for the review!
>  --noriko
>  ------------------------------------------------------------------------
>
>  --
>  389-devel mailing list
>  389-devel(a)redhat.com
>  https://www.redhat.com/mailman/listinfo/fedora-directory-devel
>