Linux Ethernet Bridging - May 2023 - [Bridge] [PATCH net-next 3/5] flow_offload: Reject matching on layer 2 miss

On Thu, May 18, 2023 at 02:33:26PM +0300, Ido Schimmel
wrote:
> Adjust drivers that support the 'FLOW_DISSECTOR_KEY_META' key to
reject
> filters that try to match on the newly added layer 2 miss option. Add an
> extack message to clearly communicate the failure reason to user space.
Hi Ido,

FLOW_DISSECTOR_KEY_META is also used in the following.
Perhaps they don't need updating. But perhaps it is worth mentioning why.

 * drivers/net/ethernet/mediatek/mtk_ppe_offload.c
 * drivers/net/ethernet/netronome/nfp/flower/conntrack.c
> 
> Example:
> 
>  # tc filter add dev swp1 egress pref 1 proto all flower skip_sw l2_miss
true action drop
>  Error: mlxsw_spectrum: Can't match on "l2_miss".
>  We have an error talking to the kernel
> 
> Acked-by: Elad Nachman <enachman at marvell.com>
> Signed-off-by: Ido Schimmel <idosch at nvidia.com>
...

On Fri, May 19, 2023 at 01:33:00PM +0200, Simon Horman
wrote:
> On Thu, May 18, 2023 at 02:33:26PM +0300, Ido Schimmel wrote:
> > Adjust drivers that support the 'FLOW_DISSECTOR_KEY_META' key
to reject
> > filters that try to match on the newly added layer 2 miss option. Add
an
> > extack message to clearly communicate the failure reason to user
space.
> 
> Hi Ido,
> 
> FLOW_DISSECTOR_KEY_META is also used in the following.
> Perhaps they don't need updating. But perhaps it is worth mentioning
why.
Good point.
> 
>  * drivers/net/ethernet/mediatek/mtk_ppe_offload.c
This driver does not seem to do anything with this key. TBH, I'm not
sure what is the purpose of this hunk:

if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_META)) {
	struct flow_match_meta match;

	flow_rule_match_meta(rule, &match);
} else {
	return -EOPNOTSUPP;
}

Felix, can you comment?
Original patch:
https://lore.kernel.org/netdev/20230518113328.1952135-4-idosch at nvidia.com/
>  * drivers/net/ethernet/netronome/nfp/flower/conntrack.c
My understanding is that this code is for netfilter offload (not tc)
which does not use the new bit. Adding a check would therefore be dead
code. I don't mind adding a check or mentioning in the commit message
why I didn't add one. Let me know what you prefer.

Thanks