netfilter: nftables: allow re-computing sctp CRC-32C in 'payload' statements
Subsystems (2)
Florian Westphal
maintainerPATCH 1/1 netfilter: nftables: allow re-computing sctp CRC-32C in 'payload' statements
From: Davide Caratti <[email protected]>
nftables payload statements are used to mangle SCTP headers, but they can only replace the Internet Checksum. As a consequence, nftables rules that mangle sport/dport/vtag in SCTP headers potentially generate packets that are discarded by the receiver, unless the CRC-32C is "offloaded" (e.g the rule mangles a skb having 'ip_summed' equal to 'CHECKSUM_PARTIAL'.
Fix this extending uAPI definitions and L4 checksum update function, in a way that userspace programs (e.g. nft) can instruct the kernel to compute CRC-32C in SCTP headers. Also ensure that LIBCRC32C is built if NF_TABLES is 'y' or 'm' in the kernel build configuration.
---
NB: I accidentally pushed the userspace side to nftables.git already, my bad. Here is the kernel part.
include/uapi/linux/netfilter/nf_tables.h | 2 ++ net/netfilter/Kconfig | 1 + net/netfilter/nft_payload.c | 28 ++++++++++++++++++++++++ 3 files changed, 31 insertions(+)
include/uapi/linux/netfilter/nf_tables.h +2 -0
@@ -749,10 +749,12 @@ enum nft_payload_bases {* * @NFT_PAYLOAD_CSUM_NONE: no checksumming * @NFT_PAYLOAD_CSUM_INET: internet checksum (RFC 791) +* @NFT_PAYLOAD_CSUM_SCTP: CRC-32c, for use in SCTP header (RFC 3309) */ enum nft_payload_csum_types {NFT_PAYLOAD_CSUM_NONE ,NFT_PAYLOAD_CSUM_INET , +NFT_PAYLOAD_CSUM_SCTP , };enum nft_payload_csum_flags {
net/netfilter/Kconfig +1 -0
@@ -441,6 +441,7 @@ endif # NF_CONNTRACK config NF_TABLES select NETFILTER_NETLINK + select LIBCRC32C tristate "Netfilter nf_tables support" help nftables is the new packet classification framework that intends to
net/netfilter/nft_payload.c +28 -0
@@ -22,6 +22,7 @@#include <linux/icmpv6.h> #include <linux/ip.h> #include <linux/ipv6.h> +#include <net/sctp/checksum.h> static bool nft_payload_rebuild_vlan_hdr (const struct sk_buff *skb,int mac_off,struct vlan_ethhdr *veth)
@@ -484,6 +485,19 @@ static int nft_payload_l4csum_offset(const struct nft_pktinfo *pkt,return 0 ; } +static int nft_payload_csum_sctp (struct sk_buff *skb,int offset) +{ +struct sctphdr *sh; + +if (skb_ensure_writable (skb, offset +sizeof (*sh))) +return -1 ; + + sh = (struct sctphdr *)(skb->data + offset); + sh->checksum =sctp_compute_cksum (skb, offset); + skb->ip_summed =CHECKSUM_UNNECESSARY ; +return 0 ; +} +static int nft_payload_l4csum_update (const struct nft_pktinfo *pkt,struct sk_buff *skb,__wsum fsum,__wsum tsum)
@@ -587,6 +601,13 @@ static void nft_payload_set_eval(const struct nft_expr *expr, skb_store_bits(skb, offset, src, priv->len) < 0) goto err; +if (priv->csum_type ==NFT_PAYLOAD_CSUM_SCTP && + pkt->tprot ==IPPROTO_SCTP && + skb->ip_summed !=CHECKSUM_PARTIAL ) { +if (nft_payload_csum_sctp (skb, pkt->xt.thoff)) + goto err; + } +return ; err: regs->verdict.code =NFT_BREAK ;
@@ -623,6 +644,13 @@ static int nft_payload_set_init(const struct nft_ctx *ctx,case NFT_PAYLOAD_CSUM_NONE :case NFT_PAYLOAD_CSUM_INET :break ; +case NFT_PAYLOAD_CSUM_SCTP : +if (priv->base !=NFT_PAYLOAD_TRANSPORT_HEADER ) +return -EINVAL ; + +if (priv->csum_offset != offsetof(struct sctphdr , checksum)) +return -EINVAL ; +break ;default :return -EOPNOTSUPP ; }
Signature · 1 line
2.26.2
Pablo Neira Ayuso
maintainerQuoted 8 lines · Florian Westphal
From: Davide Caratti <[email protected]>
nftables payload statements are used to mangle SCTP headers, but they can only replace the Internet Checksum. As a consequence, nftables rules that mangle sport/dport/vtag in SCTP headers potentially generate packets that are discarded by the receiver, unless the CRC-32C is "offloaded" (e.g the rule mangles a skb having '
ip_summed' equal to 'CHECKSUM_PARTIAL'.
Quoted 4 lines
Fix this extending uAPI definitions and L4 checksum update function, in a way that userspace programs (e.g. nft) can instruct the kernel to compute CRC-32C in SCTP headers. Also ensure that LIBCRC32C is built if
NF_TABLESis 'y' or 'm' in the kernel build configuration.
I have just passed up this to net-next to help improve chances this hits upstream in this merge window.
Thanks.