commit 868d523535c2 ("bpf: add bpf_skb_adjust_room encap flags") introduced support to bpf_skb_adjust_room for GSO-friendly GRE and UDP encapsulation. For GSO to work for skbs, the inner headers (mac and network) need to be marked. For L3 encapsulation using bpf_skb_adjust_room, the mac and network headers are identical. Here we provide a way of specifying the inner mac header length for cases where L2 encap is desired. Such an approach can support encapsulated ethernet headers, MPLS headers etc. For example to convert from a packet of form [eth][ip][tcp] to [eth][ip][udp][inner mac][ip][tcp], something like the following could be done: headroom = sizeof(iph) + sizeof(struct udphdr) + inner_maclen; ret = bpf_skb_adjust_room(skb, headroom, BPF_ADJ_ROOM_MAC, BPF_F_ADJ_ROOM_ENCAP_L4_UDP | BPF_F_ADJ_ROOM_ENCAP_L3_IPV4 | BPF_F_ADJ_ROOM_ENCAP_L2(inner_maclen));
Unified · Split
| @@ -1523,6 +1523,10 @@ union bpf_attr { · 1522 unmodified lines | ||
| 1523 | 1523 | * * **BPF_F_ADJ_ROOM_ENCAP_L4_UDP **: |
| 1524 | 1524 | * Use with ENCAP_L3 flags to further specify the tunnel type. |
| 1525 | 1525 | * |
| 1526 | + * * **BPF_F_ADJ_ROOM_ENCAP_L2(len) **: | |
| 1527 | + * Use with ENCAP_L3/L4 flags to further specify the tunnel | |
| 1528 | + * type; **len** is the length of the inner MAC header. | |
| 1529 | + * | |
| 1526 | 1530 | * A call to this helper is susceptible to change the underlaying |
| 1527 | 1531 | * packet buffer. Therefore, at load time, all checks on pointers |
| 1528 | 1532 | * previously done by the verifier are invalidated and must be |
| @@ -2664,10 +2668,16 @@ enum bpf_func_id { · 1135 unmodified lines | ||
| 2664 | 2668 | /* BPF_FUNC_skb_adjust_room flags. */ |
| 2665 | 2669 | #define BPF_F_ADJ_ROOM_FIXED_GSO (1ULL << 0) |
| 2666 | 2670 | |
| 2671 | +#define BPF_ADJ_ROOM_ENCAP_L2_MASK 0xff | |
| 2672 | +#define BPF_ADJ_ROOM_ENCAP_L2_SHIFT 56 | |
| 2673 | + | |
| 2667 | 2674 | #define BPF_F_ADJ_ROOM_ENCAP_L3_IPV4 (1ULL << 1) |
| 2668 | 2675 | #define BPF_F_ADJ_ROOM_ENCAP_L3_IPV6 (1ULL << 2) |
| 2669 | 2676 | #define BPF_F_ADJ_ROOM_ENCAP_L4_GRE (1ULL << 3) |
| 2670 | 2677 | #define BPF_F_ADJ_ROOM_ENCAP_L4_UDP (1ULL << 4) |
| 2678 | +#define BPF_F_ADJ_ROOM_ENCAP_L2(len) (((__u64)len & \ | |
| 2679 | + BPF_ADJ_ROOM_ENCAP_L2_MASK) \ | |
| 2680 | + << BPF_ADJ_ROOM_ENCAP_L2_SHIFT) | |
| 2671 | 2681 | |
| 2672 | 2682 | /* Mode for BPF_FUNC_skb_adjust_room helper. */ |
| 2673 | 2683 | enum bpf_adj_room_mode { |
| @@ -2969,11 +2969,14 @@ static u32 bpf_skb_net_base_len(const struct sk_buff *skb) · 2968 unmodified lines | ||
| 2969 | 2969 | #define BPF_F_ADJ_ROOM_MASK (BPF_F_ADJ_ROOM_FIXED_GSO | \ |
| 2970 | 2970 | BPF_F_ADJ_ROOM_ENCAP_L3_MASK | \ |
| 2971 | 2971 | BPF_F_ADJ_ROOM_ENCAP_L4_GRE | \ |
| 2972 | − BPF_F_ADJ_ROOM_ENCAP_L4_UDP) | |
| 2972 | + BPF_F_ADJ_ROOM_ENCAP_L4_UDP | \ | |
| 2973 | + BPF_F_ADJ_ROOM_ENCAP_L2( \ | |
| 2974 | + BPF_ADJ_ROOM_ENCAP_L2_MASK)) | |
| 2973 | 2975 | |
| 2974 | 2976 | static int bpf_skb_net_grow(struct sk_buff *skb, u32 off, u32 len_diff, |
| 2975 | 2977 | u64 flags) |
| 2976 | 2978 | { |
| 2979 | + u8 inner_mac_len = flags >> BPF_ADJ_ROOM_ENCAP_L2_SHIFT; | |
| 2977 | 2980 | bool encap = flags & BPF_F_ADJ_ROOM_ENCAP_L3_MASK; |
| 2978 | 2981 | u16 mac_len = 0, inner_net = 0, inner_trans = 0; |
| 2979 | 2982 | unsigned int gso_type = SKB_GSO_DODGY; |
| @@ -3008,6 +3011,8 @@ static int bpf_skb_net_grow(struct sk_buff *skb, u32 off, u32 len_diff, · 28 unmodified lines | ||
| 3008 | 3011 | |
| 3009 | 3012 | mac_len = skb->network_header - skb->mac_header; |
| 3010 | 3013 | inner_net = skb->network_header; |
| 3014 | + if (inner_mac_len > len_diff) | |
| 3015 | + return -EINVAL; | |
| 3011 | 3016 | inner_trans = skb->transport_header; |
| 3012 | 3017 | } |
| 3013 | 3018 | |
| @@ -3016,8 +3021,7 @@ static int bpf_skb_net_grow(struct sk_buff *skb, u32 off, u32 len_diff, · 2 unmodified lines | ||
| 3016 | 3021 | return ret; |
| 3017 | 3022 | |
| 3018 | 3023 | if (encap) { |
| 3019 | − /* inner mac == inner_net on l3 encap */ | |
| 3020 | − skb->inner_mac_header = inner_net; | |
| 3024 | + skb->inner_mac_header = inner_net - inner_mac_len; | |
| 3021 | 3025 | skb->inner_network_header = inner_net; |
| 3022 | 3026 | skb->inner_transport_header = inner_trans; |
| 3023 | 3027 | skb_set_inner_protocol(skb, skb->protocol); |
| @@ -3031,7 +3035,7 @@ static int bpf_skb_net_grow(struct sk_buff *skb, u32 off, u32 len_diff, · 7 unmodified lines | ||
| 3031 | 3035 | gso_type |= SKB_GSO_GRE; |
| 3032 | 3036 | else if (flags & BPF_F_ADJ_ROOM_ENCAP_L3_IPV6) |
| 3033 | 3037 | gso_type |= SKB_GSO_IPXIP6; |
| 3034 | − else | |
| 3038 | + else if (flags & BPF_F_ADJ_ROOM_ENCAP_L3_IPV4) | |
| 3035 | 3039 | gso_type |= SKB_GSO_IPXIP4; |
| 3036 | 3040 | |
| 3037 | 3041 | if (flags & BPF_F_ADJ_ROOM_ENCAP_L4_GRE || |