netdev - Re: [PATCH 1/2] bpf: add table ID to bpf_fib

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <d8184124-3ff6-9c3c-07cf-78407738d616@meta.com>
Date: Fri, 26 May 2023 09:58:19 -0700
From: Yonghong Song <yhs@...a.com>
To: Louis DeLosSantos <louis.delos.devel@...il.com>,
        John Fastabend <john.fastabend@...il.com>, bpf@...r.kernel.org,
        netdev@...r.kernel.org
Cc: Martin KaFai Lau <martin.lau@...ux.dev>,
        Stanislav Fomichev <sdf@...gle.com>, razor@...ckwall.org
Subject: Re: [PATCH 1/2] bpf: add table ID to bpf_fib_lookup BPF helper



On 5/26/23 7:07 AM, Louis DeLosSantos wrote:
> On Thu, May 25, 2023 at 11:48:12PM -0700, John Fastabend wrote:
>> Louis DeLosSantos wrote:
>>> Add ability to specify routing table ID to the `bpf_fib_lookup` BPF
>>> helper.
>>>
>>> A new field `tbid` is added to `struct bpf_fib_lookup` used as
>>> parameters to the `bpf_fib_lookup` BPF helper.
>>>
>>> When the helper is called with the `BPF_FIB_LOOKUP_DIRECT` flag and the
>>> `tbid` field in `struct bpf_fib_lookup` is greater then 0, the `tbid`
>>> field will be used as the table ID for the fib lookup.
>>>
>>> If the `tbid` does not exist the fib lookup will fail with
>>> `BPF_FIB_LKUP_RET_NOT_FWDED`.
>>>
>>> The `tbid` field becomes a union over the vlan related output fields in
>>> `struct bpf_fib_lookup` and will be zeroed immediately after usage.
>>>
>>> This functionality is useful in containerized environments.
>>>
>>> For instance, if a CNI wants to dictate the next-hop for traffic leaving
>>> a container it can create a container-specific routing table and perform
>>> a fib lookup against this table in a "host-net-namespace-side" TC program.
>>>
>>> This functionality also allows `ip rule` like functionality at the TC
>>> layer, allowing an eBPF program to pick a routing table based on some
>>> aspect of the sk_buff.
>>>
>>> As a concrete use case, this feature will be used in Cilium's SRv6 L3VPN
>>> datapath.
>>>
>>> When egress traffic leaves a Pod an eBPF program attached by Cilium will
>>> determine which VRF the egress traffic should target, and then perform a
>>> FIB lookup in a specific table representing this VRF's FIB.
>>>
>>> Signed-off-by: Louis DeLosSantos <louis.delos.devel@...il.com>
>>> ---
>>>   include/uapi/linux/bpf.h       | 17 ++++++++++++++---
>>>   net/core/filter.c              | 12 ++++++++++++
>>>   tools/include/uapi/linux/bpf.h | 17 ++++++++++++++---
>>>   3 files changed, 40 insertions(+), 6 deletions(-)
>>>
>>
>> Looks good one question. Should we hide tbid behind a flag we have
>> lots of room. Is there any concern a user could feed a bpf_fib_lookup
>> into the helper without clearing the vlan fields? Perhaps by
>> pulling the struct from a map or something where it had been
>> previously used.
>>
>> Thanks,
>> John
> 
> This is a fair point.
> 
> I could imagine a scenario where an individual is caching bpf_fib_lookup structs,
> pulls in a kernel with this change, and is now accidentally feeding the stale vlan
> fields as table ID's, since their code is using `BPF_FIB_LOOKUP_DIRECT` with
> the old semantics.
> 
> Guarding with a new flag like this (just a quick example, not a full diff)...
> 
> ```
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index 2096fbb328a9b..22095ccaaa64d 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -6823,6 +6823,7 @@ enum {
>          BPF_FIB_LOOKUP_DIRECT  = (1U << 0),
>          BPF_FIB_LOOKUP_OUTPUT  = (1U << 1),
>          BPF_FIB_LOOKUP_SKIP_NEIGH = (1U << 2),
> +       BPF_FIB_LOOKUP_TBID    = (1U << 3),
>   };
>   
>   enum {
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 6f710aa0a54b3..9b78460e39af2 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -5803,7 +5803,7 @@ static int bpf_ipv4_fib_lookup(struct net *net, struct bpf_fib_lookup *params,
>                  u32 tbid = l3mdev_fib_table_rcu(dev) ? : RT_TABLE_MAIN;
>                  struct fib_table *tb;
>   
> -               if (params->tbid) {
> +               if (flags & BPF_FIB_LOOKUP_TBID) {
>                          tbid = params->tbid;
>                          /* zero out for vlan output */
>                          params->tbid = 0;
> ```
> 
> Maybe a bit safer, you're right.
> 
> In this case the semantics around `BPF_FIB_LOOKUP_DIRECT` remain exactly the same,
> and if we do `flags = BPF_FIB_LOOKUP_DIRECT | BPF_FIB_LOOKUP_TBID`, only then will
> the `tbid` field in the incoming params wil be considered.
> 
> If I squint at this, it technically also allows us to consider `tbid=0` as a
> valid table id, since the caller now explicitly opts into it, where previously
> table id 0 was not selectable, tho I don't know if there's a *real* use case
> for selecting the `all` table.
> 
> I'm happy to make this change, what are your thoughts?

Sounds good to me so we won't reject legal table id.

>