On 9 Apr 2021, at 2:56, John Fastabend wrote:

Lorenzo Bianconi wrote:
This series introduce XDP multi-buffer support. The mvneta driver is
the first to support these new "non-linear" xdp_{buff,frame}. Reviewers
please focus on how these new types of xdp_{buff,frame} packets
traverse the different layers and the layout design. It is on purpose
that BPF-helpers are kept simple, as we don't want to expose the
internal layout to allow later changes.

For now, to keep the design simple and to maintain performance, the XDP
BPF-prog (still) only have access to the first-buffer. It is left for
later (another patchset) to add payload access across multiple buffers. This patchset should still allow for these future extensions. The goal
is to lift the XDP MTU restriction that comes with XDP, but maintain
same performance as before.

The main idea for the new multi-buffer layout is to reuse the same
layout used for non-linear SKB. We introduced a "xdp_shared_info" data structure at the end of the first buffer to link together subsequent buffers. xdp_shared_info will alias skb_shared_info allowing to keep most of the frags in the same cache-line (while with skb_shared_info only the first fragment will be placed in the first "shared_info" cache-line). Moreover we introduced some
xdp_shared_info helpers aligned to skb_frag* ones.
Converting xdp_frame to SKB and deliver it to the network stack is shown in patch 07/14. Building the SKB, the xdp_shared_info structure will be converted
in a skb_shared_info one.

A multi-buffer bit (mb) has been introduced in xdp_{buff,frame} structure to notify the bpf/network layer if this is a xdp multi-buffer frame (mb = 1)
or not (mb = 0).
The mb bit will be set by a xdp multi-buffer capable driver only for
non-linear frames maintaining the capability to receive linear frames
without any extra cost since the xdp_shared_info structure at the end
of the first buffer will be initialized only if mb is set.

Typical use cases for this series are:
- Jumbo-frames
- Packet header split (please see Google���s use-case @ NetDevConf 0x14, [0])

A new frame_length field has been introduce in XDP ctx in order to notify the
eBPF layer about the total frame size (linear + paged parts).

bpf_xdp_adjust_tail and bpf_xdp_copy helpers have been modified to take into
account xdp multi-buff frames.

I just read the commit messages for v8 so far. But, I'm still wondering how
to handle use cases where we want to put extra bytes at the end of the
packet, or really anywhere in the general case. We can extend tail with above
is there anyway to then write into that extra space?

I think most use cases will only want headers so we can likely make it
a callout to a helper. Could we add something like, xdp_get_bytes(start, end)
to pull in the bytes?

My dumb pseudoprogram being something like,

  trailer[16] = {0,1,2,3,4,5,6,7,8,9,a,b,c,d,e}
  trailer_size = 16;
  old_end = xdp->length;
  new_end = xdp->length + trailer_size;

  err = bpf_xdp_adjust_tail(xdp, trailer_size)
  if (err) return err;

  err = xdp_get_bytes(xdp, old_end, new_end);
  if (err) return err;

  memcpy(xdp->data, trailer, trailer_size);

Do you think that could work if we code up xdp_get_bytes()? Does the driver have enough context to adjust xdp to map to my get_bytes() call? I think
so but we should check.

I was thinking of doing something like the below, but I have no cycles to work on it:

void *bpf_xdp_access_bytes(struct xdp_buff *xdp_md, u32 offset, int *len, void *buffer)
This function returns a pointer to the packet data, which can be
             accessed linearly for a maximum of *len* bytes.

*offset* marks the starting point in the packet for which you
             would like to get a data pointer.

*len* point to an initialized integer which tells the helper how many bytes from *offset* you would like to access. Supplying a value of 0 or less will tell the helper to report back how many bytes are available linearly from the offset (in this case the value of *buffer* is ignored). On return, the helper will
             update this value with the length available to access
             linearly at the address returned.

*buffer* point to an optional buffer which MUST be the same size as *\*len* and will be used to copy in the data if it's not
             available linearly.

Returns a pointer to the packet data requested accessible with
             a maximum length of *\*len*. NULL is returned on failure.

Note that if a *buffer* is supplied and the data is not available
             linearly, the content is copied. In this case a pointer to
             *buffer* is returned.

int bpf_xdp_store_bytes(struct xdp_buff *xdp_md, u32 offset, const void *from, u32 len)
Store *len* bytes from address *from* into the packet associated to *xdp_md*, at *offset*. This function will take care of copying
             data to multi-buffer XDP packets.

A call to this helper is susceptible to change the underlying packet buffer. Therefore, at load time, all checks on pointers previously done by the verifier are invalidated and must be
             performed again, if the helper is used in combination with
             direct packet access.

             0 on success, or a negative error in case of failure.

More info about the main idea behind this approach can be found here [1][2].

Thanks for working on this!

Reply via email to