Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 2428 bytes --]

On Thu, Aug 22, 2024 at 11:31:37AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> We encounter a tcp drop issue in our cloud environment. Packet GROed in
> host forwards to a VM virtio_net nic with net_failover enabled. VM acts
> as a IPVS LB with ipip encapsulation. The full path like:
> host gro -> vm virtio_net rx -> net_failover rx -> ipvs fullnat
>  -> ipip encap -> net_failover tx -> virtio_net tx
> 
> When net_failover transmits a ipip pkt (gso_type = 0x0103, which means
> SKB_GSO_TCPV4, SKB_GSO_DODGY and SKB_GSO_IPXIP4), there is no gso
> did because it supports TSO and GSO_IPXIP4. But network_header points to
> inner ip header.
> 
> Call Trace:
>  tcp4_gso_segment        ------> return NULL
>  inet_gso_segment        ------> inner iph, network_header points to
>  ipip_gso_segment
>  inet_gso_segment        ------> outer iph
>  skb_mac_gso_segment

> Afterwards virtio_net transmits the pkt, only inner ip header is modified.
> And the outer one just keeps unchanged. The pkt will be dropped in remote
> host.

That may appear like a transient connection issue or permanently
impossible connection?

> Call Trace:
>  inet_gso_segment        ------> inner iph, outer iph is skipped
>  skb_mac_gso_segment
>  __skb_gso_segment
>  validate_xmit_skb
>  validate_xmit_skb_list
>  sch_direct_xmit
>  __qdisc_run
>  __dev_queue_xmit        ------> virtio_net
>  dev_hard_start_xmit
>  __dev_queue_xmit        ------> net_failover
>  ip_finish_output2
>  ip_output
>  iptunnel_xmit
>  ip_tunnel_xmit
>  ipip_tunnel_xmit        ------> ipip
>  dev_hard_start_xmit
>  __dev_queue_xmit
>  ip_finish_output2
>  ip_output
>  ip_forward
>  ip_rcv
>  __netif_receive_skb_one_core
>  netif_receive_skb_internal
>  napi_gro_receive
>  receive_buf
>  virtnet_poll
>  net_rx_action
> 
> The root cause of this issue is specific with the rare combination of
> SKB_GSO_DODGY and a tunnel device that adds an SKB_GSO_ tunnel option.
> SKB_GSO_DODGY is set from external virtio_net. We need to reset network
> header when callbacks.gso_segment() returns NULL.

Who's in control of these configuration (who can cause this incorrect
packet being sent)?

> This patch also includes ipv6_gso_segment(), considering SIT, etc.
> 
> The Linux kernel CVE team has assigned CVE-2022-48936 to this issue.

What is the security issue here?

Thanks,
Michal

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Greg Kroah-Hartman]

On Tue, Aug 27, 2024 at 05:02:36PM +0200, Michal Koutný wrote:
> On Thu, Aug 22, 2024 at 11:31:37AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > We encounter a tcp drop issue in our cloud environment. Packet GROed in
> > host forwards to a VM virtio_net nic with net_failover enabled. VM acts
> > as a IPVS LB with ipip encapsulation. The full path like:
> > host gro -> vm virtio_net rx -> net_failover rx -> ipvs fullnat
> >  -> ipip encap -> net_failover tx -> virtio_net tx
> > 
> > When net_failover transmits a ipip pkt (gso_type = 0x0103, which means
> > SKB_GSO_TCPV4, SKB_GSO_DODGY and SKB_GSO_IPXIP4), there is no gso
> > did because it supports TSO and GSO_IPXIP4. But network_header points to
> > inner ip header.
> > 
> > Call Trace:
> >  tcp4_gso_segment        ------> return NULL
> >  inet_gso_segment        ------> inner iph, network_header points to
> >  ipip_gso_segment
> >  inet_gso_segment        ------> outer iph
> >  skb_mac_gso_segment
> 
> > Afterwards virtio_net transmits the pkt, only inner ip header is modified.
> > And the outer one just keeps unchanged. The pkt will be dropped in remote
> > host.
> 
> That may appear like a transient connection issue or permanently
> impossible connection?

I don't know.

> > Call Trace:
> >  inet_gso_segment        ------> inner iph, outer iph is skipped
> >  skb_mac_gso_segment
> >  __skb_gso_segment
> >  validate_xmit_skb
> >  validate_xmit_skb_list
> >  sch_direct_xmit
> >  __qdisc_run
> >  __dev_queue_xmit        ------> virtio_net
> >  dev_hard_start_xmit
> >  __dev_queue_xmit        ------> net_failover
> >  ip_finish_output2
> >  ip_output
> >  iptunnel_xmit
> >  ip_tunnel_xmit
> >  ipip_tunnel_xmit        ------> ipip
> >  dev_hard_start_xmit
> >  __dev_queue_xmit
> >  ip_finish_output2
> >  ip_output
> >  ip_forward
> >  ip_rcv
> >  __netif_receive_skb_one_core
> >  netif_receive_skb_internal
> >  napi_gro_receive
> >  receive_buf
> >  virtnet_poll
> >  net_rx_action
> > 
> > The root cause of this issue is specific with the rare combination of
> > SKB_GSO_DODGY and a tunnel device that adds an SKB_GSO_ tunnel option.
> > SKB_GSO_DODGY is set from external virtio_net. We need to reset network
> > header when callbacks.gso_segment() returns NULL.
> 
> Who's in control of these configuration (who can cause this incorrect
> packet being sent)?
> 
> > This patch also includes ipv6_gso_segment(), considering SIT, etc.
> > 
> > The Linux kernel CVE team has assigned CVE-2022-48936 to this issue.
> 
> What is the security issue here?

This was assigned as part of the import of the Linux kernel GSD entries
into CVEs as required by the CVE board of directors (hence the 2022
date).  If you don't feel this should be assigned a CVE, just let me
know and I will be glad to reject it.

thanks,

greg k-h

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Michal Koutný]

[-- Attachment #1: Type: text/plain, Size: 571 bytes --]

On Wed, Aug 28, 2024 at 09:30:08AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > What is the security issue here?
> 
> This was assigned as part of the import of the Linux kernel GSD entries
> into CVEs as required by the CVE board of directors (hence the 2022
> date).  If you don't feel this should be assigned a CVE, just let me
> know and I will be glad to reject it.

The address of original author bounces back. Willem, could you please
help explaining context of the change? (~the questions in my previous
message).

Thanks,
Michal

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Willem de Bruijn]

On Thu, Aug 29, 2024 at 12:18 PM Michal Koutný <mkoutny@suse.com> wrote:
>
> On Wed, Aug 28, 2024 at 09:30:08AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > > What is the security issue here?
> >
> > This was assigned as part of the import of the Linux kernel GSD entries
> > into CVEs as required by the CVE board of directors (hence the 2022
> > date).  If you don't feel this should be assigned a CVE, just let me
> > know and I will be glad to reject it.
>
> The address of original author bounces back. Willem, could you please
> help explaining context of the change? (~the questions in my previous
> message).

I don't know why this has a CVE.

The patch reports that the negative effect is a drop due to a corrupted packet.

According to the CVE report this requires both user input with
virtio_net_hdr, which is privileged, and a tunnel device configured,
which again is privileged.

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Greg Kroah-Hartman]

On Thu, Aug 29, 2024 at 12:53:34PM -0400, Willem de Bruijn wrote:
> On Thu, Aug 29, 2024 at 12:18 PM Michal Koutný <mkoutny@suse.com> wrote:
> >
> > On Wed, Aug 28, 2024 at 09:30:08AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > > > What is the security issue here?
> > >
> > > This was assigned as part of the import of the Linux kernel GSD entries
> > > into CVEs as required by the CVE board of directors (hence the 2022
> > > date).  If you don't feel this should be assigned a CVE, just let me
> > > know and I will be glad to reject it.
> >
> > The address of original author bounces back. Willem, could you please
> > help explaining context of the change? (~the questions in my previous
> > message).
> 
> I don't know why this has a CVE.
> 
> The patch reports that the negative effect is a drop due to a corrupted packet.
> 
> According to the CVE report this requires both user input with
> virtio_net_hdr, which is privileged, and a tunnel device configured,
> which again is privileged.
> 

Ok, should it be rejected then?  If so, just let me know.

thanks,

greg k-h

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Willem de Bruijn]

On Thu, Aug 29, 2024 at 12:58 PM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> On Thu, Aug 29, 2024 at 12:53:34PM -0400, Willem de Bruijn wrote:
> > On Thu, Aug 29, 2024 at 12:18 PM Michal Koutný <mkoutny@suse.com> wrote:
> > >
> > > On Wed, Aug 28, 2024 at 09:30:08AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > > > > What is the security issue here?
> > > >
> > > > This was assigned as part of the import of the Linux kernel GSD entries
> > > > into CVEs as required by the CVE board of directors (hence the 2022
> > > > date).  If you don't feel this should be assigned a CVE, just let me
> > > > know and I will be glad to reject it.
> > >
> > > The address of original author bounces back. Willem, could you please
> > > help explaining context of the change? (~the questions in my previous
> > > message).
> >
> > I don't know why this has a CVE.
> >
> > The patch reports that the negative effect is a drop due to a corrupted packet.
> >
> > According to the CVE report this requires both user input with
> > virtio_net_hdr, which is privileged, and a tunnel device configured,
> > which again is privileged.
> >
>
> Ok, should it be rejected then?  If so, just let me know.

It is a legitimate bug fix, definitely stable material.

With the fix backported to all these branches, not sure what, if
anything, more is needed wrt the CVE.

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Greg Kroah-Hartman]

On Thu, Aug 29, 2024 at 01:07:36PM -0400, Willem de Bruijn wrote:
> On Thu, Aug 29, 2024 at 12:58 PM Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
> >
> > On Thu, Aug 29, 2024 at 12:53:34PM -0400, Willem de Bruijn wrote:
> > > On Thu, Aug 29, 2024 at 12:18 PM Michal Koutný <mkoutny@suse.com> wrote:
> > > >
> > > > On Wed, Aug 28, 2024 at 09:30:08AM GMT, Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:
> > > > > > What is the security issue here?
> > > > >
> > > > > This was assigned as part of the import of the Linux kernel GSD entries
> > > > > into CVEs as required by the CVE board of directors (hence the 2022
> > > > > date).  If you don't feel this should be assigned a CVE, just let me
> > > > > know and I will be glad to reject it.
> > > >
> > > > The address of original author bounces back. Willem, could you please
> > > > help explaining context of the change? (~the questions in my previous
> > > > message).
> > >
> > > I don't know why this has a CVE.
> > >
> > > The patch reports that the negative effect is a drop due to a corrupted packet.
> > >
> > > According to the CVE report this requires both user input with
> > > virtio_net_hdr, which is privileged, and a tunnel device configured,
> > > which again is privileged.
> > >
> >
> > Ok, should it be rejected then?  If so, just let me know.
> 
> It is a legitimate bug fix, definitely stable material.
> 
> With the fix backported to all these branches, not sure what, if
> anything, more is needed wrt the CVE.

Ok, for now I'll go revoke this as that feels the safest thing to do at
the moment.

The "frankenkernel" distros can decide if they want to pick this up or
not, everyone sane has had it for years now :)

thanks,

greg k-h

Re: CVE-2022-48936: gso: do not skip outer ip header in case of ipip and net_failover

[Michal Koutný]

On Thu, Aug 29, 2024 at 01:07:36PM GMT, Willem de Bruijn <willemb@google.com> wrote:
> With the fix backported to all these branches, not sure what, if
> anything, more is needed wrt the CVE.

My goal is to maintain some positive signal/noise in the assigned CVEs.

(Also, there are other kernel branches to evaluate as Greg says :-))

On Thu, Aug 29, 2024 at 12:53:34PM GMT, Willem de Bruijn <willemb@google.com> wrote:
> The patch reports that the negative effect is a drop due to a corrupted packet.
> 
> According to the CVE report this requires both user input with
> virtio_net_hdr, which is privileged, and a tunnel device configured,
> which again is privileged.

(Unless inside netns, I assume. Though...)
...that affects only scope of the same user, right? (The effect doesn't
cross boundary of any security domains.)

So far there is nothing in my (incomplete) understadning that would
classify this as a vulnerability (bugfix it is indeed).

Thanks,
Michal