On Fri, 30 Oct 2020 12:40:21 -0700
John Fastabend [off-list ref] wrote:

Jesper Dangaard Brouer wrote:

quoted

The BPF-helpers for FIB lookup (bpf_xdp_fib_lookup and bpf_skb_fib_lookup)
can perform MTU check and return BPF_FIB_LKUP_RET_FRAG_NEEDED.  The BPF-prog
don't know the MTU value that caused this rejection.

If the BPF-prog wants to implement PMTU (Path MTU Discovery) (rfc1191) it
need to know this MTU value for the ICMP packet.

Patch change lookup and result struct bpf_fib_lookup, to contain this MTU
value as output via a union with 'tot_len' as this is the value used for
the MTU lookup.

V5:
 - Fixed uninit value spotted by Dan Carpenter.
 - Name struct output member mtu_result

Reported-by: kernel test robot <redacted>
Reported-by: Dan Carpenter <redacted>
Signed-off-by: Jesper Dangaard Brouer <redacted>
---
 include/uapi/linux/bpf.h       |   11 +++++++++--
 net/core/filter.c              |   22 +++++++++++++++-------
 tools/include/uapi/linux/bpf.h |   11 +++++++++--
 3 files changed, 33 insertions(+), 11 deletions(-)

diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index e6ceac3f7d62..01b2b17c645a 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h

@@ -2219,6 +2219,9 @@ union bpf_attr {
  *		* > 0 one of **BPF_FIB_LKUP_RET_** codes explaining why the
  *		  packet is not forwarded or needs assist from full stack
  *
+ *		If lookup fails with BPF_FIB_LKUP_RET_FRAG_NEEDED, then the MTU
+ *		was exceeded and result params->mtu contains the MTU.
+ *  

Do we need to hide this behind a flag? It seems otherwise you might confuse
users. I imagine on error we could reuse the params arg, but now we changed
the tot_len value underneath them?

The principle behind this bpf_fib_lookup helper, is that params (struct
bpf_fib_lookup) is used for both input and output (results). Almost
every field is change after the lookup. (For performance reasons this
is kept at 64 bytes (cache-line))  Thus, users of this helper already
expect/knows the contents of params have changed.

-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer


struct bpf_fib_lookup {
	/* input:  network family for lookup (AF_INET, AF_INET6)
	 * output: network family of egress nexthop
	 */
	__u8	family;

	/* set if lookup is to consider L4 data - e.g., FIB rules */
	__u8	l4_protocol;
	__be16	sport;
	__be16	dport;

	union {	/* used for MTU check */
		/* input to lookup */
		__u16	tot_len; /* total length of packet from network hdr */

		/* output: MTU value (if requested check_mtu) */
		__u16	mtu_result;
	};
	/* input: L3 device index for lookup
	 * output: device index from FIB lookup
	 */
	__u32	ifindex;

	union {
		/* inputs to lookup */
		__u8	tos;		/* AF_INET  */
		__be32	flowinfo;	/* AF_INET6, flow_label + priority */

		/* output: metric of fib result (IPv4/IPv6 only) */
		__u32	rt_metric;
	};

	union {
		__be32		ipv4_src;
		__u32		ipv6_src[4];  /* in6_addr; network order */
	};

	/* input to bpf_fib_lookup, ipv{4,6}_dst is destination address in
	 * network header. output: bpf_fib_lookup sets to gateway address
	 * if FIB lookup returns gateway route
	 */
	union {
		__be32		ipv4_dst;
		__u32		ipv6_dst[4];  /* in6_addr; network order */
	};

	/* output */
	__be16	h_vlan_proto;
	__be16	h_vlan_TCI;
	__u8	smac[6];     /* ETH_ALEN */
	__u8	dmac[6];     /* ETH_ALEN */
};