Re: KASAN: use-after-free Read in __xfrm6_tunnel_spi_lookup

[Dmitry Vyukov]

On Fri, Sep 11, 2020 at 10:06 AM Steffen Klassert
<steffen.klassert@xxxxxxxxxxx> wrote:
>
> On Thu, Sep 10, 2020 at 10:09:50AM +0200, Dmitry Vyukov wrote:
> > On Thu, Sep 10, 2020 at 10:08 AM B K Karthik <bkkarthik@xxxxxxxxxxxx> wrote:
> > >
> > > On Thu, Sep 10, 2020 at 1:32 PM Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:
> > > >
> > > > On Thu, Sep 10, 2020 at 9:20 AM Anant Thazhemadam
> > > > <anant.thazhemadam@xxxxxxxxx> wrote:
> > > > > Looks like this bug is no longer valid. I'm not sure which commit seems to have fixed it. Can this be marked as invalid or closed yet?
> > > >
> > > > You can see on the dashboard (or in mailing list archives) that B K
> > > > Karthik tested a patch for this bug in July:
> > > > https://syzkaller.appspot.com/bug?extid=72ff2fa98097767b5a27
> > > >
> > > > So perhaps that patch fixes it? Karthik, did you send it? Was it
> > > > merged? Did the commit include the syzbot Reported-by tag?
> > > >
> > >
> > > I did send it. I was taking a u32 spi value and casting it to a
> > > pointer to an IP address. Steffen Klassert
> > > <steffen.klassert@xxxxxxxxxxx> pointed out to me that the approach i
> > > was looking at was completely wrong.
> > > https://lkml.org/lkml/2020/7/27/361 is the conversation. hope this
> > > helps.
> >
> > +Steffen, was there any other fix merged for this?
>
> I think that was already fixed before the sysbot report came in by
> commit 8b404f46dd6a ("xfrm: interface: not xfrmi_ipv6/ipip_handler twice")

Thanks

Let's close the report then:

#syz fix: xfrm: interface: not xfrmi_ipv6/ipip_handler twice