On Thu, 28 Jan 2021 22:51:23 -0800
John Fastabend <john.fastabend@xxxxxxxxx> wrote:
> Jesper Dangaard Brouer wrote:
> > This BPF-helper bpf_check_mtu() works for both XDP and TC-BPF programs.
> >
> > The SKB object is complex and the skb->len value (accessible from
> > BPF-prog) also include the length of any extra GRO/GSO segments, but
> > without taking into account that these GRO/GSO segments get added
> > transport (L4) and network (L3) headers before being transmitted. Thus,
> > this BPF-helper is created such that the BPF-programmer don't need to
> > handle these details in the BPF-prog.
> >
> > The API is designed to help the BPF-programmer, that want to do packet
> > context size changes, which involves other helpers. These other helpers
> > usually does a delta size adjustment. This helper also support a delta
> > size (len_diff), which allow BPF-programmer to reuse arguments needed by
> > these other helpers, and perform the MTU check prior to doing any actual
> > size adjustment of the packet context.
> >
> > It is on purpose, that we allow the len adjustment to become a negative
> > result, that will pass the MTU check. This might seem weird, but it's not
> > this helpers responsibility to "catch" wrong len_diff adjustments. Other
> > helpers will take care of these checks, if BPF-programmer chooses to do
> > actual size adjustment.
The nitpick below about len adjust can become negative, is on purpose
and why is described in above.
> >
> > V13:
> > - Enforce flag BPF_MTU_CHK_SEGS cannot use len_diff.
> >
> > V12:
> > - Simplify segment check that calls skb_gso_validate_network_len.
> > - Helpers should return long
> >
> > V9:
> > - Use dev->hard_header_len (instead of ETH_HLEN)
> > - Annotate with unlikely req from Daniel
> > - Fix logic error using skb_gso_validate_network_len from Daniel
> >
> > V6:
> > - Took John's advice and dropped BPF_MTU_CHK_RELAX
> > - Returned MTU is kept at L3-level (like fib_lookup)
> >
> > V4: Lot of changes
> > - ifindex 0 now use current netdev for MTU lookup
> > - rename helper from bpf_mtu_check to bpf_check_mtu
> > - fix bug for GSO pkt length (as skb->len is total len)
> > - remove __bpf_len_adj_positive, simply allow negative len adj
Notice V4 comment about "allow negative len adj"
> > Signed-off-by: Jesper Dangaard Brouer <brouer@xxxxxxxxxx>
> > ---
> > include/uapi/linux/bpf.h | 67 ++++++++++++++++++++++++
> > net/core/filter.c | 114 ++++++++++++++++++++++++++++++++++++++++
> > tools/include/uapi/linux/bpf.h | 67 ++++++++++++++++++++++++
> > 3 files changed, 248 insertions(+)
> >
> > diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> > index 05bfc8c843dc..f17381a337ec 100644
> > --- a/include/uapi/linux/bpf.h
> > +++ b/include/uapi/linux/bpf.h
> > @@ -3839,6 +3839,61 @@ union bpf_attr {
>
> [...]
>
> > +
> > +BPF_CALL_5(bpf_skb_check_mtu, struct sk_buff *, skb,
> > + u32, ifindex, u32 *, mtu_len, s32, len_diff, u64, flags)
>
> Maybe worth mentioning in description we expect len_diff < skb->len,
> at least I expect that otherwise result may be undefined.
>
> > +{
> > + int ret = BPF_MTU_CHK_RET_FRAG_NEEDED;
> > + struct net_device *dev = skb->dev;
> > + int skb_len, dev_len;
> > + int mtu;
>
> Perhaps getting a bit nit-picky here but shouldn't skb_len, dev_len
> and mtu all be 'unsigned int'
>
> Then all the types will align. I guess MTUs are small so it
> doesn't really matter, but is easier to read IMO.
We need signed types, this is a deliberate choice made based on
discussion in V4.
> > +
> > + if (unlikely(flags & ~(BPF_MTU_CHK_SEGS)))
> > + return -EINVAL;
> > +
> > + if (unlikely(flags & BPF_MTU_CHK_SEGS && len_diff))
> > + return -EINVAL;
> > +
> > + dev = __dev_via_ifindex(dev, ifindex);
> > + if (unlikely(!dev))
> > + return -ENODEV;
> > +
> > + mtu = READ_ONCE(dev->mtu);
> > +
> > + dev_len = mtu + dev->hard_header_len;
> > + skb_len = skb->len + len_diff; /* minus result pass check */
> > + if (skb_len <= dev_len) {
>
> If skb_len is unsigned it will be >> dev_len when skb->len < len_diff. I
> think its a good idea to throw an error if skb_len calculation goes
> negative?
No, as comment says /* minus result pass check */.
And explained in patch desc.
> > + ret = BPF_MTU_CHK_RET_SUCCESS;
> > + goto out;
> > + }
> > + /* At this point, skb->len exceed MTU, but as it include length of all
> > + * segments, it can still be below MTU. The SKB can possibly get
> > + * re-segmented in transmit path (see validate_xmit_skb). Thus, user
> > + * must choose if segs are to be MTU checked.
> > + */
> > + if (skb_is_gso(skb)) {
> > + ret = BPF_MTU_CHK_RET_SUCCESS;
> > +
> > + if (flags & BPF_MTU_CHK_SEGS &&
> > + !skb_gso_validate_network_len(skb, mtu))
> > + ret = BPF_MTU_CHK_RET_SEGS_TOOBIG;
> > + }
> > +out:
> > + /* BPF verifier guarantees valid pointer */
> > + *mtu_len = mtu;
> > +
> > + return ret;
> > +}
> > +
> > +BPF_CALL_5(bpf_xdp_check_mtu, struct xdp_buff *, xdp,
> > + u32, ifindex, u32 *, mtu_len, s32, len_diff, u64, flags)
> > +{
> > + struct net_device *dev = xdp->rxq->dev;
> > + int xdp_len = xdp->data_end - xdp->data;
> > + int ret = BPF_MTU_CHK_RET_SUCCESS;
> > + int mtu, dev_len;
>
> Same comment about types.
>
> > +
> > + /* XDP variant doesn't support multi-buffer segment check (yet) */
> > + if (unlikely(flags))
> > + return -EINVAL;
> > +
> > + dev = __dev_via_ifindex(dev, ifindex);
> > + if (unlikely(!dev))
> > + return -ENODEV;
> > +
> > + mtu = READ_ONCE(dev->mtu);
> > +
> > + /* Add L2-header as dev MTU is L3 size */
> > + dev_len = mtu + dev->hard_header_len;
> > +
> > + xdp_len += len_diff; /* minus result pass check */
> > + if (xdp_len > dev_len)
> > + ret = BPF_MTU_CHK_RET_FRAG_NEEDED;
> > +
> > + /* BPF verifier guarantees valid pointer */
> > + *mtu_len = mtu;
> > +
> > + return ret;
> > +}
>
> Otherwise LGTM.
Thanks
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
