| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <87blkvmsd3.fsf@toke.dk>
Date: Sat, 04 Jul 2020 13:28:40 +0200
From: Toke Høiland-Jørgensen <toke@...hat.com>
To: Daniel Borkmann <daniel@...earbox.net>
Cc: davem@...emloft.net, netdev@...r.kernel.org, bpf@...r.kernel.org,
cake@...ts.bufferbloat.net, Davide Caratti <dcaratti@...hat.com>,
Jiri Pirko <jiri@...nulli.us>,
Jamal Hadi Salim <jhs@...atatu.com>,
Cong Wang <xiyou.wangcong@...il.com>,
Roman Mashak <mrv@...atatu.com>,
Lawrence Brakmo <brakmo@...com>,
Ilya Ponetayev <i.ponetaev@...systems.com>, kafai@...com,
alexei.starovoitov@...il.com, edumazet@...gle.com
Subject: Re: [PATCH net v3] sched: consistently handle layer3 header accesses in the presence of VLANs
Daniel Borkmann <daniel@...earbox.net> writes:
> On 7/3/20 10:26 PM, Toke Høiland-Jørgensen wrote:
>> There are a couple of places in net/sched/ that check skb->protocol and act
>> on the value there. However, in the presence of VLAN tags, the value stored
>> in skb->protocol can be inconsistent based on whether VLAN acceleration is
>> enabled. The commit quoted in the Fixes tag below fixed the users of
>> skb->protocol to use a helper that will always see the VLAN ethertype.
>>
>> However, most of the callers don't actually handle the VLAN ethertype, but
>> expect to find the IP header type in the protocol field. This means that
>> things like changing the ECN field, or parsing diffserv values, stops
>> working if there's a VLAN tag, or if there are multiple nested VLAN
>> tags (QinQ).
>>
>> To fix this, change the helper to take an argument that indicates whether
>> the caller wants to skip the VLAN tags or not. When skipping VLAN tags, we
>> make sure to skip all of them, so behaviour is consistent even in QinQ
>> mode.
>>
>> To make the helper usable from the ECN code, move it to if_vlan.h instead
>> of pkt_sched.h.
>>
>> v3:
>> - Remove empty lines
>> - Move vlan variable definitions inside loop in skb_protocol()
>> - Also use skb_protocol() helper in IP{,6}_ECN_decapsulate() and
>> bpf_skb_ecn_set_ce()
>>
>> v2:
>> - Use eth_type_vlan() helper in skb_protocol()
>> - Also fix code that reads skb->protocol directly
>> - Change a couple of 'if/else if' statements to switch constructs to avoid
>> calling the helper twice
>>
>> Reported-by: Ilya Ponetayev <i.ponetaev@...systems.com>
>> Fixes: d8b9605d2697 ("net: sched: fix skb->protocol use in case of accelerated vlan path")
>> Signed-off-by: Toke Høiland-Jørgensen <toke@...hat.com>
>> ---
>> include/linux/if_vlan.h | 28 ++++++++++++++++++++++++++++
>> include/net/inet_ecn.h | 25 +++++++++++++++++--------
>> include/net/pkt_sched.h | 11 -----------
>> net/core/filter.c | 10 +++++++---
>> net/sched/act_connmark.c | 9 ++++++---
>> net/sched/act_csum.c | 2 +-
>> net/sched/act_ct.c | 9 ++++-----
>> net/sched/act_ctinfo.c | 9 ++++++---
>> net/sched/act_mpls.c | 2 +-
>> net/sched/act_skbedit.c | 2 +-
>> net/sched/cls_api.c | 2 +-
>> net/sched/cls_flow.c | 8 ++++----
>> net/sched/cls_flower.c | 2 +-
>> net/sched/em_ipset.c | 2 +-
>> net/sched/em_ipt.c | 2 +-
>> net/sched/em_meta.c | 2 +-
>> net/sched/sch_cake.c | 4 ++--
>> net/sched/sch_dsmark.c | 6 +++---
>> net/sched/sch_teql.c | 2 +-
>> 19 files changed, 86 insertions(+), 51 deletions(-)
>>
>> diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h
>> index b05e855f1ddd..427a5b8597c2 100644
>> --- a/include/linux/if_vlan.h
>> +++ b/include/linux/if_vlan.h
>> @@ -308,6 +308,34 @@ static inline bool eth_type_vlan(__be16 ethertype)
>> }
>> }
>>
>> +/* A getter for the SKB protocol field which will handle VLAN tags consistently
>> + * whether VLAN acceleration is enabled or not.
>> + */
>> +static inline __be16 skb_protocol(const struct sk_buff *skb, bool skip_vlan)
>> +{
>> + unsigned int offset = skb_mac_offset(skb) + sizeof(struct ethhdr);
>> + __be16 proto = skb->protocol;
>> +
>> + if (!skip_vlan)
>> + /* VLAN acceleration strips the VLAN header from the skb and
>> + * moves it to skb->vlan_proto
>> + */
>> + return skb_vlan_tag_present(skb) ? skb->vlan_proto : proto;
>> +
>> + while (eth_type_vlan(proto)) {
>> + struct vlan_hdr vhdr, *vh;
>> +
>> + vh = skb_header_pointer(skb, offset, sizeof(vhdr), &vhdr);
>> + if (!vh)
>> + break;
>> +
>> + proto = vh->h_vlan_encapsulated_proto;
>> + offset += sizeof(vhdr);
>> + }
>
> Hm, why is the while loop 'unbounded'? Does it even make sense to have
> a packet with hundreds of vlan hdrs in there what you'd end up
> walking? What if an attacker crafts a max sized packet with only
> vlan_hdr forcing exorbitant looping in fast-path here (e.g. via
> af_packet)?
Hmm, I guess you're right that could theoretically happen. But on the
other hand, a lot of drivers seem to be cheerfully calling
vlan_get_protocol() on incoming packets, which doesn't have a limit on
the depth either.
I guess I could add a depth limit, but in that case I suppose that
should also be added to vlan_get_protocol() (or the two should be
consolidated). WDYT?
> Did you validate that skb_mac_offset() is always valid for the
> call-sites you converted? (We have a skb_mac_header_was_set() test to
> probe for whether skb->mac_header is set to ~0.)
Not extensively; I kinda assumed it would always be valid at those call
sites, since the callers go on to call ip_hdr() or something similar
right afterwards.
I guess Toshiaki's suggestion to use vlan_get_protocol() could be a way
around this, as that seems to deal with skb->mac_len being 0.
-Toke
Powered by blists - more mailing lists