This updates the existing unaligned_access_hacks patch to reduce unalignment traps to nearly zero for ipv4 and ipv6, in the normal path, in encapsulation, for native apps, and in the qdiscs.
Raw IPv6 performance over ethernet more than doubled. For more detail, see: http://www.bufferbloat.net/projects/cerowrt/issues/360 A thx to Felix Feitkau, Ketan Kulkarni, Robert Bradley. --- .../patches-3.3/902-unaligned_access_hacks.patch | 603 ++++++++++++++++++-- 1 files changed, 568 insertions(+), 35 deletions(-) diff --git a/target/linux/ar71xx/patches-3.3/902-unaligned_access_hacks.patch b/target/linux/ar71xx/patches-3.3/902-unaligned_access_hacks.patch index 921cf19..7d1ba42 100644 --- a/target/linux/ar71xx/patches-3.3/902-unaligned_access_hacks.patch +++ b/target/linux/ar71xx/patches-3.3/902-unaligned_access_hacks.patch @@ -1,3 +1,60 @@ +From edcc5edc0f9038001d2d7ff000e256b46c644570 Mon Sep 17 00:00:00 2001 +From: Dave Taht <dave.t...@bufferbloat.net> +Date: Sun, 22 Apr 2012 14:16:00 -0700 +Subject: [PATCH] Comprehensive ag71xx ipv4 and ipv6 network alignment fixes + +The ag71xx driver can only transport data on an unaligned +boundry on some platforms. + +The original alignment fixes patch addressed common +ipv4 issues only. + +It did not address: + +ipv6 +encapsulated packets +queueing disciplines and traffic shaping issues +tcp timestamped packets and tcp connections +or vpns + +With this version of the patch, ipv6 performance more +than doubled, ipv4 improved by over 12% in the general case, +queuing disciplines no longer threw traps like crazy, +local apps like web proxies and network benchmarks +improved, and the router felt much "smoother" under heavy load. + +IPv6 before: 110Mbit +IPv6 after: 252Mbit +IPv4 before: 220Mbit +Ipv4 after: 254Mbit +--- + arch/mips/include/asm/checksum.h | 46 ++++++++++++++++++++++-- + include/linux/icmpv6.h | 2 +- + include/linux/if_vlan.h | 2 +- + include/linux/igmp.h | 8 ++-- + include/linux/in6.h | 8 ++-- + include/linux/ip.h | 10 +++--- + include/linux/ipv6.h | 4 +- + include/linux/tcp.h | 7 ++-- + include/net/flow_keys.h | 2 +- + include/net/ip6_checksum.h | 5 +++ + include/net/ndisc.h | 6 ++-- + net/core/flow_dissector.c | 7 +++- + net/core/secure_seq.c | 3 +- + net/ipv4/netfilter/nf_conntrack_l3proto_ipv4.c | 5 ++- + net/ipv4/tcp.c | 2 +- + net/ipv4/tcp_input.c | 6 ++-- + net/ipv6/af_inet6.c | 2 +- + net/ipv6/datagram.c | 16 ++++---- + net/ipv6/exthdrs.c | 2 +- + net/ipv6/netfilter/ip6t_LOG.c | 4 +- + net/ipv6/route.c | 2 +- + net/ipv6/xfrm6_policy.c | 4 +- + net/xfrm/xfrm_input.c | 7 +++- + 23 files changed, 107 insertions(+), 53 deletions(-) + +diff --git a/arch/mips/include/asm/checksum.h b/arch/mips/include/asm/checksum.h +index f2f7c6c..dce734c 100644 --- a/arch/mips/include/asm/checksum.h +++ b/arch/mips/include/asm/checksum.h @@ -12,6 +12,7 @@ @@ -8,48 +65,185 @@ #include <asm/uaccess.h> -@@ -104,26 +105,30 @@ static inline __sum16 ip_fast_csum(const - const unsigned int *stop = word + ihl; - unsigned int csum; - int carry; +@@ -98,6 +99,41 @@ static inline __sum16 csum_fold(__wsum sum) + * By Jorge Cwik <jo...@laser.satlink.net>, adapted for linux by + * Arnt Gulbrandsen. + */ ++static inline __sum16 ip_fast_csum_unaligned(const void *iph, unsigned int ihl) ++{ ++ const unsigned int *word = iph; ++ const unsigned int *stop = word + ihl; ++ unsigned int csum; ++ int carry; + unsigned int w; - -- csum = word[0]; -- csum += word[1]; -- carry = (csum < word[1]); ++ + csum = __get_unaligned_cpu32(word++); + + w = __get_unaligned_cpu32(word++); + csum += w; + carry = (csum < w); - csum += carry; - -- csum += word[2]; -- carry = (csum < word[2]); ++ csum += carry; ++ + w = __get_unaligned_cpu32(word++); + csum += w; + carry = (csum < w); - csum += carry; - -- csum += word[3]; -- carry = (csum < word[3]); ++ csum += carry; ++ + w = __get_unaligned_cpu32(word++); + csum += w; + carry = (csum < w); - csum += carry; - -- word += 4; - do { -- csum += *word; -- carry = (csum < *word); ++ csum += carry; ++ ++ do { + w = __get_unaligned_cpu32(word++); + csum += w; + carry = (csum < w); - csum += carry; -- word++; - } while (word != stop); ++ csum += carry; ++ } while (word != stop); ++ ++ return csum_fold(csum); ++} ++ + static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl) + { + const unsigned int *word = iph; +@@ -105,6 +141,9 @@ static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl) + unsigned int csum; + int carry; + ++ if ((unsigned int) iph & 3) ++ return ip_fast_csum_unaligned(iph,ihl); ++ + csum = word[0]; + csum += word[1]; + carry = (csum < word[1]); +@@ -192,14 +231,15 @@ static inline __sum16 ip_compute_csum(const void *buff, int len) + return csum_fold(csum_partial(buff, len, 0)); + } + +-#define _HAVE_ARCH_IPV6_CSUM +-static __inline__ __sum16 csum_ipv6_magic(const struct in6_addr *saddr, ++#define _HAVE_ARCH_IPV6_CSUM_ALIGNED ++static __inline__ __sum16 csum_ipv6_magic_aligned( ++ const struct in6_addr *saddr, + const struct in6_addr *daddr, + __u32 len, unsigned short proto, + __wsum sum) + { + __asm__( +- " .set push # csum_ipv6_magic\n" ++ " .set push # csum_ipv6_magic_aligned\n" + " .set noreorder \n" + " .set noat \n" + " addu %0, %5 # proto (long in network byte order)\n" +diff --git a/include/linux/icmpv6.h b/include/linux/icmpv6.h +index ba45e6b..5bb0258 100644 +--- a/include/linux/icmpv6.h ++++ b/include/linux/icmpv6.h +@@ -76,7 +76,7 @@ struct icmp6hdr { + #define icmp6_addrconf_other icmp6_dataun.u_nd_ra.other + #define icmp6_rt_lifetime icmp6_dataun.u_nd_ra.rt_lifetime + #define icmp6_router_pref icmp6_dataun.u_nd_ra.router_pref +-}; ++} __packed; + + #ifdef __KERNEL__ + #include <linux/skbuff.h> +diff --git a/include/linux/if_vlan.h b/include/linux/if_vlan.h +index 13aff1e..add341d 100644 +--- a/include/linux/if_vlan.h ++++ b/include/linux/if_vlan.h +@@ -39,7 +39,7 @@ + struct vlan_hdr { + __be16 h_vlan_TCI; + __be16 h_vlan_encapsulated_proto; +-}; ++} __packed; + + /** + * struct vlan_ethhdr - vlan ethernet header (ethhdr + vlan_hdr) +diff --git a/include/linux/igmp.h b/include/linux/igmp.h +index 82de336..a17b50c 100644 +--- a/include/linux/igmp.h ++++ b/include/linux/igmp.h +@@ -32,7 +32,7 @@ struct igmphdr { + __u8 code; /* For newer IGMP */ + __sum16 csum; + __be32 group; +-}; ++} __packed; + + /* V3 group record types [grec_type] */ + #define IGMPV3_MODE_IS_INCLUDE 1 +@@ -48,7 +48,7 @@ struct igmpv3_grec { + __be16 grec_nsrcs; + __be32 grec_mca; + __be32 grec_src[0]; +-}; ++} __packed; + + struct igmpv3_report { + __u8 type; +@@ -57,7 +57,7 @@ struct igmpv3_report { + __be16 resv2; + __be16 ngrec; + struct igmpv3_grec grec[0]; +-}; ++} __packed; - return csum_fold(csum); + struct igmpv3_query { + __u8 type; +@@ -78,7 +78,7 @@ struct igmpv3_query { + __u8 qqic; + __be16 nsrcs; + __be32 srcs[0]; +-}; ++} __packed; + + #define IGMP_HOST_MEMBERSHIP_QUERY 0x11 /* From RFC1112 */ + #define IGMP_HOST_MEMBERSHIP_REPORT 0x12 /* Ditto */ +diff --git a/include/linux/in6.h b/include/linux/in6.h +index 097a34b..0d0f788 100644 +--- a/include/linux/in6.h ++++ b/include/linux/in6.h +@@ -36,7 +36,7 @@ struct in6_addr { + #define s6_addr in6_u.u6_addr8 + #define s6_addr16 in6_u.u6_addr16 + #define s6_addr32 in6_u.u6_addr32 +-}; ++} __packed; + + /* IPv6 Wildcard Address (::) and Loopback Address (::1) defined in RFC2553 + * NOTE: Be aware the IN6ADDR_* constants and in6addr_* externals are defined +@@ -61,7 +61,7 @@ struct sockaddr_in6 { + __be32 sin6_flowinfo; /* IPv6 flow information */ + struct in6_addr sin6_addr; /* IPv6 address */ + __u32 sin6_scope_id; /* scope id (new in RFC2553) */ +-}; ++} __packed; + + struct ipv6_mreq { + /* IPv6 multicast address of group */ +@@ -69,7 +69,7 @@ struct ipv6_mreq { + + /* local IPv6 address of interface */ + int ipv6mr_ifindex; +-}; ++} __packed; + + #define ipv6mr_acaddr ipv6mr_multiaddr + +@@ -83,7 +83,7 @@ struct in6_flowlabel_req { + __u16 flr_linger; + __u32 __flr_pad; + /* Options in format of IPV6_PKTOPTIONS */ +-}; ++} __packed; + + #define IPV6_FL_A_GET 0 + #define IPV6_FL_A_PUT 1 +diff --git a/include/linux/ip.h b/include/linux/ip.h +index bd0a2a8..85ad985 100644 --- a/include/linux/ip.h +++ b/include/linux/ip.h @@ -102,7 +102,7 @@ struct iphdr { @@ -61,8 +255,49 @@ #ifdef __KERNEL__ #include <linux/skbuff.h> +@@ -125,25 +125,25 @@ struct ip_auth_hdr { + __be32 spi; + __be32 seq_no; /* Sequence number */ + __u8 auth_data[0]; /* Variable len but >=4. Mind the 64 bit alignment! */ +-}; ++} __packed; + + struct ip_esp_hdr { + __be32 spi; + __be32 seq_no; /* Sequence number */ + __u8 enc_data[0]; /* Variable len but >=8. Mind the 64 bit alignment! */ +-}; ++} __packed; + + struct ip_comp_hdr { + __u8 nexthdr; + __u8 flags; + __be16 cpi; +-}; ++} __packed; + + struct ip_beet_phdr { + __u8 nexthdr; + __u8 hdrlen; + __u8 padlen; + __u8 reserved; +-}; ++} __packed; + + #endif /* _LINUX_IP_H */ +diff --git a/include/linux/ipv6.h b/include/linux/ipv6.h +index 6318268..07f35fa 100644 --- a/include/linux/ipv6.h +++ b/include/linux/ipv6.h +@@ -49,7 +49,7 @@ struct ipv6_rt_hdr { + * type specific data + * variable length field + */ +-}; ++} __packed; + + + struct ipv6_opt_hdr { @@ -126,7 +126,7 @@ struct ipv6hdr { struct in6_addr saddr; @@ -72,6 +307,8 @@ #ifdef __KERNEL__ /* +diff --git a/include/linux/tcp.h b/include/linux/tcp.h +index 3c7ffdb..680e0e8 100644 --- a/include/linux/tcp.h +++ b/include/linux/tcp.h @@ -54,7 +54,7 @@ struct tcphdr { @@ -83,17 +320,122 @@ /* * The union cast uses a gcc extension to avoid aliasing problems ---- a/include/linux/udp.h -+++ b/include/linux/udp.h -@@ -24,7 +24,7 @@ struct udphdr { - __be16 dest; - __be16 len; - __sum16 check; +@@ -64,9 +64,10 @@ struct tcphdr { + union tcp_word_hdr { + struct tcphdr hdr; + __be32 words[5]; +-}; ++} __packed; + +-#define tcp_flag_word(tp) ( ((union tcp_word_hdr *)(tp))->words [3]) ++#define tcp_flag_word2(tp) ( ((union tcp_word_hdr *)(tp))->words [3]) ++#define tcp_flag_word(tp) ( __get_unaligned_cpu32(&(((union tcp_word_hdr *)(tp))->words [3]))) + + enum { + TCP_FLAG_CWR = __cpu_to_be32(0x00800000), +diff --git a/include/net/flow_keys.h b/include/net/flow_keys.h +index 80461c1..ffd6fbd 100644 +--- a/include/net/flow_keys.h ++++ b/include/net/flow_keys.h +@@ -10,7 +10,7 @@ struct flow_keys { + __be16 port16[2]; + }; + u8 ip_proto; +-}; ++} __packed; + + extern bool skb_flow_dissect(const struct sk_buff *skb, struct flow_keys *flow); + #endif +diff --git a/include/net/ip6_checksum.h b/include/net/ip6_checksum.h +index bc1b0fd..caef125 100644 +--- a/include/net/ip6_checksum.h ++++ b/include/net/ip6_checksum.h +@@ -45,6 +45,11 @@ static __inline__ __sum16 csum_ipv6_magic(const struct in6_addr *saddr, + __u32 uproto; + __u32 sum = (__force u32)csum; + ++#ifdef _HAVE_ARCH_IPV6_CSUM_ALIGNED ++ if (((((unsigned int) saddr | (unsigned int) daddr)) & 3) == 0) ++ return csum_ipv6_magic_aligned(saddr,daddr,len,proto,csum); ++#endif ++ + sum += (__force u32)saddr->s6_addr32[0]; + carry = (sum < (__force u32)saddr->s6_addr32[0]); + sum += carry; +diff --git a/include/net/ndisc.h b/include/net/ndisc.h +index a9d350e..5be936d 100644 +--- a/include/net/ndisc.h ++++ b/include/net/ndisc.h +@@ -62,18 +62,18 @@ struct nd_msg { + struct icmp6hdr icmph; + struct in6_addr target; + __u8 opt[0]; +-}; ++} __packed; + + struct rs_msg { + struct icmp6hdr icmph; + __u8 opt[0]; +-}; ++} __packed; + + struct ra_msg { + struct icmp6hdr icmph; + __be32 reachable_time; + __be32 retrans_timer; -}; +} __packed; - /* UDP socket options */ - #define UDP_CORK 1 /* Never send partially complete segments */ + struct nd_opt_hdr { + __u8 nd_opt_type; +diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c +index a225089..019970f 100644 +--- a/net/core/flow_dissector.c ++++ b/net/core/flow_dissector.c +@@ -17,7 +17,10 @@ static void iph_to_flow_copy_addrs(struct flow_keys *flow, const struct iphdr *i + { + BUILD_BUG_ON(offsetof(typeof(*flow), dst) != + offsetof(typeof(*flow), src) + sizeof(flow->src)); +- memcpy(&flow->src, &iph->saddr, sizeof(flow->src) + sizeof(flow->dst)); ++ /* memcpy(&flow->src, &iph->saddr, sizeof(flow->src) + sizeof(flow->dst)); */ ++ flow->src = iph->saddr; ++ flow->dst = iph->daddr; ++ + } + + bool skb_flow_dissect(const struct sk_buff *skb, struct flow_keys *flow) +@@ -136,7 +139,7 @@ ipv6: + nhoff += poff; + ports = skb_header_pointer(skb, nhoff, sizeof(_ports), &_ports); + if (ports) +- flow->ports = *ports; ++ flow->ports = __get_unaligned_cpu32(ports); + } + + return true; +diff --git a/net/core/secure_seq.c b/net/core/secure_seq.c +index 99b2596..2127551 100644 +--- a/net/core/secure_seq.c ++++ b/net/core/secure_seq.c +@@ -7,6 +7,7 @@ + #include <linux/hrtimer.h> + #include <linux/ktime.h> + #include <linux/string.h> ++#include <linux/unaligned/packed_struct.h> + + #include <net/secure_seq.h> + +@@ -46,7 +47,7 @@ __u32 secure_tcpv6_sequence_number(const __be32 *saddr, const __be32 *daddr, + + memcpy(hash, saddr, 16); + for (i = 0; i < 4; i++) +- secret[i] = net_secret[i] + (__force u32)daddr[i]; ++ secret[i] = net_secret[i] + (__force u32) __get_unaligned_cpu32(&daddr[i]); + secret[4] = net_secret[4] + + (((__force u16)sport << 16) + (__force u16)dport); + for (i = 5; i < MD5_MESSAGE_BYTES / 4; i++) +diff --git a/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4.c b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4.c +index de9da21..2704407 100644 --- a/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4.c +++ b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4.c @@ -14,6 +14,7 @@ @@ -104,7 +446,7 @@ #include <net/route.h> #include <net/ip.h> -@@ -44,8 +45,8 @@ static bool ipv4_pkt_to_tuple(const stru +@@ -44,8 +45,8 @@ static bool ipv4_pkt_to_tuple(const struct sk_buff *skb, unsigned int nhoff, if (ap == NULL) return false; @@ -115,3 +457,194 @@ return true; } +diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c +index 22ef5f9..bf3f773 100644 +--- a/net/ipv4/tcp.c ++++ b/net/ipv4/tcp.c +@@ -2834,7 +2834,7 @@ found: + + p = *head; + th2 = tcp_hdr(p); +- tcp_flag_word(th2) |= flags & (TCP_FLAG_FIN | TCP_FLAG_PSH); ++ tcp_flag_word2(th2) = tcp_flag_word(th2) | flags & (TCP_FLAG_FIN | TCP_FLAG_PSH); + + out_check_final: + flush = len < mss; +diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c +index b5e315f..1a0e649 100644 +--- a/net/ipv4/tcp_input.c ++++ b/net/ipv4/tcp_input.c +@@ -3944,13 +3944,13 @@ static int tcp_parse_aligned_timestamp(struct tcp_sock *tp, const struct tcphdr + { + const __be32 *ptr = (const __be32 *)(th + 1); + +- if (*ptr == htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) ++ if (__get_unaligned_cpu32(ptr) == htonl((TCPOPT_NOP << 24) | (TCPOPT_NOP << 16) + | (TCPOPT_TIMESTAMP << 8) | TCPOLEN_TIMESTAMP)) { + tp->rx_opt.saw_tstamp = 1; + ++ptr; +- tp->rx_opt.rcv_tsval = ntohl(*ptr); ++ tp->rx_opt.rcv_tsval = ntohl(__get_unaligned_cpu32(ptr)); + ++ptr; +- tp->rx_opt.rcv_tsecr = ntohl(*ptr); ++ tp->rx_opt.rcv_tsecr = ntohl(__get_unaligned_cpu32(ptr)); + return 1; + } + return 0; +diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c +index 273f48d..c649aea 100644 +--- a/net/ipv6/af_inet6.c ++++ b/net/ipv6/af_inet6.c +@@ -695,7 +695,7 @@ int ipv6_opt_accepted(struct sock *sk, struct sk_buff *skb) + if ((opt->hop && (np->rxopt.bits.hopopts || + np->rxopt.bits.ohopopts)) || + ((IPV6_FLOWINFO_MASK & +- *(__be32 *)skb_network_header(skb)) && ++ __get_unaligned_cpu32((__be32 *)skb_network_header(skb))) && + np->rxopt.bits.rxflow) || + (opt->srcrt && (np->rxopt.bits.srcrt || + np->rxopt.bits.osrcrt)) || +diff --git a/net/ipv6/datagram.c b/net/ipv6/datagram.c +index 251e7cd..0b0abd8 100644 +--- a/net/ipv6/datagram.c ++++ b/net/ipv6/datagram.c +@@ -358,12 +358,12 @@ int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len) + *(struct in6_addr *)(nh + serr->addr_offset); + if (np->sndflow) + sin->sin6_flowinfo = +- (*(__be32 *)(nh + serr->addr_offset - 24) & ++ (__get_unaligned_cpu32((__be32 *)(nh + serr->addr_offset - 24)) & + IPV6_FLOWINFO_MASK); + if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL) + sin->sin6_scope_id = IP6CB(skb)->iif; + } else { +- ipv6_addr_set_v4mapped(*(__be32 *)(nh + serr->addr_offset), ++ ipv6_addr_set_v4mapped(__get_unaligned_cpu32((__be32 *)(nh + serr->addr_offset)), + &sin->sin6_addr); + } + } +@@ -485,12 +485,12 @@ int datagram_recv_ctl(struct sock *sk, struct msghdr *msg, struct sk_buff *skb) + } + + if (np->rxopt.bits.rxtclass) { +- int tclass = (ntohl(*(__be32 *)ipv6_hdr(skb)) >> 20) & 0xff; ++ int tclass = (ntohl(__get_unaligned_cpu32((__be32 *)ipv6_hdr(skb))) >> 20) & 0xff; + put_cmsg(msg, SOL_IPV6, IPV6_TCLASS, sizeof(tclass), &tclass); + } + +- if (np->rxopt.bits.rxflow && (*(__be32 *)nh & IPV6_FLOWINFO_MASK)) { +- __be32 flowinfo = *(__be32 *)nh & IPV6_FLOWINFO_MASK; ++ if (np->rxopt.bits.rxflow && (__get_unaligned_cpu32((__be32 *)nh) & IPV6_FLOWINFO_MASK)) { ++ __be32 flowinfo = __get_unaligned_cpu32((__be32 *)nh) & IPV6_FLOWINFO_MASK; + put_cmsg(msg, SOL_IPV6, IPV6_FLOWINFO, sizeof(flowinfo), &flowinfo); + } + +@@ -585,7 +585,7 @@ int datagram_recv_ctl(struct sock *sk, struct msghdr *msg, struct sk_buff *skb) + + sin6.sin6_family = AF_INET6; + sin6.sin6_addr = ipv6_hdr(skb)->daddr; +- sin6.sin6_port = ports[1]; ++ sin6.sin6_port = __get_unaligned_cpu16(&(ports[1])); + sin6.sin6_flowinfo = 0; + sin6.sin6_scope_id = 0; + +@@ -677,12 +677,12 @@ int datagram_send_ctl(struct net *net, struct sock *sk, + } + + if (fl6->flowlabel&IPV6_FLOWINFO_MASK) { +- if ((fl6->flowlabel^*(__be32 *)CMSG_DATA(cmsg))&~IPV6_FLOWINFO_MASK) { ++ if ((fl6->flowlabel^__get_unaligned_cpu32((__be32 *)CMSG_DATA(cmsg)))&~IPV6_FLOWINFO_MASK) { + err = -EINVAL; + goto exit_f; + } + } +- fl6->flowlabel = IPV6_FLOWINFO_MASK & *(__be32 *)CMSG_DATA(cmsg); ++ fl6->flowlabel = IPV6_FLOWINFO_MASK & __get_unaligned_cpu32((__be32 *)CMSG_DATA(cmsg)); + break; + + case IPV6_2292HOPOPTS: +diff --git a/net/ipv6/exthdrs.c b/net/ipv6/exthdrs.c +index 3d641b6..7c072da 100644 +--- a/net/ipv6/exthdrs.c ++++ b/net/ipv6/exthdrs.c +@@ -593,7 +593,7 @@ static int ipv6_hop_jumbo(struct sk_buff *skb, int optoff) + goto drop; + } + +- pkt_len = ntohl(*(__be32 *)(nh + optoff + 2)); ++ pkt_len = ntohl(__get_unaligned_cpu32((__be32 *)(nh + optoff + 2))); + if (pkt_len <= IPV6_MAXPLEN) { + IP6_INC_STATS_BH(net, ipv6_skb_idev(skb), + IPSTATS_MIB_INHDRERRORS); +diff --git a/net/ipv6/netfilter/ip6t_LOG.c b/net/ipv6/netfilter/ip6t_LOG.c +index e6af8d7..ef83fab 100644 +--- a/net/ipv6/netfilter/ip6t_LOG.c ++++ b/net/ipv6/netfilter/ip6t_LOG.c +@@ -64,9 +64,9 @@ static void dump_packet(struct sbuff *m, + /* Max length: 44 "LEN=65535 TC=255 HOPLIMIT=255 FLOWLBL=FFFFF " */ + sb_add(m, "LEN=%Zu TC=%u HOPLIMIT=%u FLOWLBL=%u ", + ntohs(ih->payload_len) + sizeof(struct ipv6hdr), +- (ntohl(*(__be32 *)ih) & 0x0ff00000) >> 20, ++ (ntohl(__get_unaligned_cpu32((__be32 *)ih)) & 0x0ff00000) >> 20, + ih->hop_limit, +- (ntohl(*(__be32 *)ih) & 0x000fffff)); ++ (ntohl(__get_unaligned_cpu32((__be32 *)ih)) & 0x000fffff)); + + fragment = 0; + ptr = ip6hoff + sizeof(struct ipv6hdr); +diff --git a/net/ipv6/route.c b/net/ipv6/route.c +index 4d5915d..e1e8bad 100644 +--- a/net/ipv6/route.c ++++ b/net/ipv6/route.c +@@ -877,7 +877,7 @@ void ip6_route_input(struct sk_buff *skb) + .flowi6_iif = skb->dev->ifindex, + .daddr = iph->daddr, + .saddr = iph->saddr, +- .flowlabel = (* (__be32 *) iph) & IPV6_FLOWINFO_MASK, ++ .flowlabel = (__get_unaligned_cpu32(iph)) & IPV6_FLOWINFO_MASK, + .flowi6_mark = skb->mark, + .flowi6_proto = iph->nexthdr, + }; +diff --git a/net/ipv6/xfrm6_policy.c b/net/ipv6/xfrm6_policy.c +index 8ea65e0..eecde59 100644 +--- a/net/ipv6/xfrm6_policy.c ++++ b/net/ipv6/xfrm6_policy.c +@@ -160,8 +160,8 @@ _decode_session6(struct sk_buff *skb, struct flowi *fl, int reverse) + pskb_may_pull(skb, nh + offset + 4 - skb->data))) { + __be16 *ports = (__be16 *)exthdr; + +- fl6->fl6_sport = ports[!!reverse]; +- fl6->fl6_dport = ports[!reverse]; ++ fl6->fl6_sport = __get_unaligned_cpu16(&ports[!!reverse]); ++ fl6->fl6_dport = __get_unaligned_cpu16(&ports[!reverse]); + } + fl6->flowi6_proto = nexthdr; + return; +diff --git a/net/xfrm/xfrm_input.c b/net/xfrm/xfrm_input.c +index 54a0dc2..e9c51b6 100644 +--- a/net/xfrm/xfrm_input.c ++++ b/net/xfrm/xfrm_input.c +@@ -52,6 +52,7 @@ int xfrm_parse_spi(struct sk_buff *skb, u8 nexthdr, __be32 *spi, __be32 *seq) + { + int offset, offset_seq; + int hlen; ++ __be32 *pspi, *pseq; + + switch (nexthdr) { + case IPPROTO_AH: +@@ -77,8 +78,10 @@ int xfrm_parse_spi(struct sk_buff *skb, u8 nexthdr, __be32 *spi, __be32 *seq) + if (!pskb_may_pull(skb, hlen)) + return -EINVAL; + +- *spi = *(__be32*)(skb_transport_header(skb) + offset); +- *seq = *(__be32*)(skb_transport_header(skb) + offset_seq); ++ pspi = (__be32*)(skb_transport_header(skb) + offset); ++ pseq = (__be32*)(skb_transport_header(skb) + offset_seq); ++ *spi = __get_unaligned_cpu32(pspi); ++ *seq = __get_unaligned_cpu32(pseq); + return 0; + } + +-- +1.7.5.4 + -- 1.7.1 _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
