[Cake] [PATCH net v3] sched: consistently handle layer3 header accesses in the presence of VLANs

[Toke Høiland-Jørgensen]

Toshiaki Makita <toshiaki.makita1 at gmail.com> writes:

> On 2020/07/04 5:26, Toke Høiland-Jørgensen wrote:
> ...
>> +/* 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);
>> +	}
>
> Why don't you use __vlan_get_protocol() here? It looks quite similar.
> Is there any problem with using that?

TBH, I completely missed that helper. It seems to have side effects,
though (pskb_may_pull()), which is one of the things the original patch
to sch_cake that initiated all of this was trying to avoid.

I guess I could just fix that, though, and switch __vlan_get_protocol()
over to using skb_header_pointer(). Will send a follow-up to do that.

Any opinion on whether it's a good idea to limit the max parse depth
while I'm at it (see Daniel's reply)?

-Toke