linux-kernel - Re: [PATCH] bpf: provide map key to BPF program after redirect

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

Message-ID: <87wmm07z9w.fsf@toke.dk>
Date: Fri, 05 Jul 2024 15:18:35 +0200
From: Toke Høiland-Jørgensen <toke@...hat.com>
To: Florian Kauer <florian.kauer@...utronix.de>, ast@...nel.org,
 daniel@...earbox.net, john.fastabend@...il.com
Cc: davem@...emloft.net, kuba@...nel.org, hawk@...nel.org,
 edumazet@...gle.com, pabeni@...hat.com, andrii@...nel.org,
 martin.lau@...ux.dev, eddyz87@...il.com, song@...nel.org,
 yonghong.song@...ux.dev, kpsingh@...nel.org, sdf@...gle.com,
 haoluo@...gle.com, jolsa@...nel.org, netdev@...r.kernel.org,
 bpf@...r.kernel.org, linux-kernel@...r.kernel.org,
 xdp-newbies@...r.kernel.org
Subject: Re: [PATCH] bpf: provide map key to BPF program after redirect

Florian Kauer <florian.kauer@...utronix.de> writes:

> On 7/5/24 13:01, Toke Høiland-Jørgensen wrote:
>> Florian Kauer <florian.kauer@...utronix.de> writes:
>> 
>>> Both DEVMAP as well as CPUMAP provide the possibility
>>> to attach BPF programs to their entries that will be
>>> executed after a redirect was performed.
>>>
>>> With BPF_F_BROADCAST it is in also possible to execute
>>> BPF programs for multiple clones of the same XDP frame
>>> which is, for example, useful for establishing redundant
>>> traffic paths by setting, for example, different VLAN tags
>>> for the replicated XDP frames.
>>>
>>> Currently, this program itself has no information about
>>> the map entry that led to its execution. While egress_ifindex
>>> can be used to get this information indirectly and can
>>> be used for path dependent processing of the replicated frames,
>>> it does not work if multiple entries share the same egress_ifindex.
>>>
>>> Therefore, extend the xdp_md struct with a map_key
>>> that contains the key of the associated map entry
>>> after performing a redirect.
>>>
>>> See
>>> https://lore.kernel.org/xdp-newbies/5eb6070c-a12e-4d4c-a9f0-a6a6fafa41d1@linutronix.de/T/#u
>>> for the discussion that led to this patch.
>>>
>>> Signed-off-by: Florian Kauer <florian.kauer@...utronix.de>
>>> ---
>>>  include/net/xdp.h        |  3 +++
>>>  include/uapi/linux/bpf.h |  2 ++
>>>  kernel/bpf/devmap.c      |  6 +++++-
>>>  net/core/filter.c        | 18 ++++++++++++++++++
>>>  4 files changed, 28 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/net/xdp.h b/include/net/xdp.h
>>> index e6770dd40c91..e70f4dfea1a2 100644
>>> --- a/include/net/xdp.h
>>> +++ b/include/net/xdp.h
>>> @@ -86,6 +86,7 @@ struct xdp_buff {
>>>  	struct xdp_txq_info *txq;
>>>  	u32 frame_sz; /* frame size to deduce data_hard_end/reserved tailroom*/
>>>  	u32 flags; /* supported values defined in xdp_buff_flags */
>>> +	u64 map_key; /* set during redirect via a map */
>>>  };
>>>  
>>>  static __always_inline bool xdp_buff_has_frags(struct xdp_buff *xdp)
>>> @@ -175,6 +176,7 @@ struct xdp_frame {
>>>  	struct net_device *dev_rx; /* used by cpumap */
>>>  	u32 frame_sz;
>>>  	u32 flags; /* supported values defined in xdp_buff_flags */
>>> +	u64 map_key; /* set during redirect via a map */
>>>  };
>> 
>> struct xdp_frame is size constrained, so we shouldn't be using precious
>> space on this. Besides, it's not information that should be carried
>> along with the packet after transmission. So let's put it into struct
>> xdp_txq_info and read it from there the same way we do for egress_ifindex :)
>
> Very reasonable, but do you really mean struct xdp_frame or xdp_buff?
> Only the latter has the xdp_txq_info?

Well, we should have the field in neither, but xdp_frame is the one that
is size constrained. Whenever a cpumap/devmap program is run (in
xdp_bq_bpf_prog_run() and dev_map_bpf_prog_run_skb()), a struct
xdp_txq_info is prepared on the stack, so you'll just need to add
setting of the new value to that...

-Toke