bpf: make netronome nfp compatible w/ bpf_xdp_adjust_tail
authorNikita V. Shirokov <tehnerd@tehnerd.com>
Wed, 18 Apr 2018 04:42:18 +0000 (21:42 -0700)
committerDaniel Borkmann <daniel@iogearbox.net>
Wed, 18 Apr 2018 21:34:16 +0000 (23:34 +0200)
w/ bpf_xdp_adjust_tail helper xdp's data_end pointer could be changed as
well (only "decrease" of pointer's location is going to be supported).
changing of this pointer will change packet's size.
for nfp driver we will just calculate packet's length unconditionally

Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Nikita V. Shirokov <tehnerd@tehnerd.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
drivers/net/ethernet/netronome/nfp/nfp_net_common.c

index 1eb6549f2a54eeca685f14d3b851bf0db170f42a..d9111c07769904a8506c3a43542a095a97905167 100644 (file)
@@ -1722,7 +1722,7 @@ static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget)
 
                        act = bpf_prog_run_xdp(xdp_prog, &xdp);
 
-                       pkt_len -= xdp.data - orig_data;
+                       pkt_len = xdp.data_end - xdp.data;
                        pkt_off += xdp.data - orig_data;
 
                        switch (act) {