Re: [PATCH v2] kcm: remove any offset before parsing messages

[Dominique Martinet <asmadeus@codewreck.org>]

David Miller wrote on Mon, Sep 17, 2018:
> > No, they can see it, so it's possible to make a KCM program that works
> > right now if you are careful (I'm not sure why the offset within bpf is
> > different from the offset in the kernel though, it looks like the bpf
> > program skips the qos part of the control buffer)
> 
> What helper is used in the BPF program to get this offset value?
> 
> (also good info to add to the commit message)

Dave defined one himself ; for a simple protocol where the offset is in
the first four bytes of the message.

The whole bpf program could look like this:

------
struct kcm_rx_msg { int full_len; int offset; };
static inline struct kcm_rx_msg *kcm_rx_msg(struct __sk_buff *skb) {
	return (struct kcm_rx_msg *)skb->cb;
}
int decode_framing(struct __sk_buff *skb) {
	return load_word(skb, kcm_rx_msg(skb)->offset);
}
------

If we go towards documenting it, adding a helper would be useful yes;
buf if we pull that becomes unnecessary.
(I'll add the example program in the commit message anyway at your
suggestion)

-- 
Dominique Martinet