[Openswan dev] [Commits] [OPENSWAN.git] Changes to ref refs/heads/master

[Thomas Geulig]

On Tue, 04 May 2010 David McCullough wrote:
> Jivin Paul Wouters lays it down ...
>
> > On Fri, 30 Apr 2010, davidm at vault.xelerance.com wrote:
> > >    Fix assertion failure when removing routes
> > >
> > >    In delete_states_by_connection:
> > >
> > >    ASSERTION FAILED at .../programs/pluto/state.c:691: sr->eroute_owner
> > > == SOS_NOBODY
> > >
> > >    kernel_interface was left set to AUTO_PICK even though USE_KLIPS had
> > > been selected by the code.  Note sure if this actually affected netkey
> > > or others but it may.
> > >
> > >    This change makes sure that, once an interface is selected,
> > > everything points to that interface being the one.
> > >
> > >    This was a subtle side affect of
> > > fdfb59b413eec432969014762ceb847ef7e5e9a4 :-)
> >
> > I don't understand. The code reads (with + for your added lines)
> >
> > switch(kern_interface) {
> >      case AUTO_PICK:
> >  	[...]
> >
> >      case USE_KLIPS:
> > +        kern_interface = USE_KLIPS;
> >  	[...]
> >
> >   #if defined(BSD_KAME)
> >       case USE_BSDKAME:
> > +       kern_interface = USE_BSDKAME;
> >  	[...]
> >
> > Aren't you just setting it to the value it must have because otherwise it
> > would never end up in the switch statement to begin with? I mean, it
> > won't hurt, but I don't think this can be fixing any assertion failed
> > that you saw.
>
> A small change at the top this funciton meant that on an openswan userland
> compiled for both netkey and klips,  but using klips,  it would drop
>  through to the switch statement with kern_interface = AUTO_PICK,  so it
>  would then go an setup the klips interface by falling through the switch
>  statement, but kern_interface was left set to AUTO_PICK, later I would see
>  the assertion failure.

Ok, my setup is klips-only, so I didn't see that.

> This was running on the head of git master completely unmodified.
>
> Obviously somewhere else in pluto is using kern_interface to make decisions
> and getting it wrong when kern_interface==AUTO_PICK.  I didn't go looking
> for the offender.  I actually found this use 'git bisect',  decided that
> there was no way that fdfb59b413eec432969014762ceb847ef7e5e9a4 could be the
> cause and then went through it all by hand again,  and still didn't believe
> it ;-)

At least one case (I had problems with) is the function delete_ipsec_sa()
in kernel.c.

Thomas