kernel 校验和实现

kernel 校验和实现

Kernel checksum implementation

1) TCP包的错误检测使用16位累加和校验. 除了TCP包本身,
TCP校验数据块还包括源IP地址,目的IP地址, TCP包长度, TCP协议号组成的12字节伪头标.

2) 校验和为16位字补码和, 数据块长度为奇数时, 数据块末尾添零处理.
校验和的计算与顺序无关, 可以从数据块开始计算, 也可以从未尾开始向前计算.

3) 为了提高计算效率, TCP包的校验和并不一次算出,
而是采用32位部分累加和(sk->csum)进行增量计算.
csum_partial()用来计算数据块的32位部分累加和, 累加和可以用csum_fold()折叠为16位校验和.
csum_partial_copy_nocheck()可在拷贝用户数据的同时计算出它的部分累加和.

4) 为了加快执行速度, csum_partial()将8个32位字分为一组用分立的指令进行32位累加,
这样可加长循环体中指令长度, 提高CPU指令流水线的效率.

5) 并不是所有的TCP包都必须校验, skb->ip_summed用来控制校验操作.
对于loopback设备的收发包, 其skb->ip_summed设为HECKSUM_UNNECESSARY, 忽略校验过程.
如果 skb->ip_summed == CHECKSUM_HW 说明TCP包本身的校验已经由硬件链路层完成，新kernel(2.6.24)已经不使用这个选项。

下面我们介绍具体实现，基于kernel-2.6.24， x86 architecture.

[函数实现]
TCP包接收校验的初始化
static __sum16 tcp_v4_checksum_init(struct sk_buff *skb)
{
    const struct iphdr *iph = ip_hdr(skb);
    //如果TCP包本身的校验已经完成
    if (skb->ip_summed == CHECKSUM_COMPLETE) {
        if (!tcp_v4_check(skb->len, iph->saddr, iph->daddr, skb->csum)) { //附加伪头进行校验
            skb->ip_summed = CHECKSUM_UNNECESSARY;
            return 0;
        }
    }
    //生成包含伪头的累加和
    skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr, skb->len, IPPROTO_TCP, 0);
    if (skb->len <= 76) {
        return __skb_checksum_complete(skb); //计算数据部分校验和
    }
    return 0;
}
    附加伪头进行校验
static inline __sum16 tcp_v4_check(int len, __be32 saddr, __be32 daddr, __wsum base)
{
    return csum_tcpudp_magic(saddr, daddr, len, IPPROTO_TCP, base);
}
static inline __sum16 csum_tcpudp_magic(__be32 saddr, __be32 daddr, unsigned short len, unsigned short proto, __wsum sum)
{
    return csum_fold(csum_tcpudp_nofold(saddr, daddr, len, proto, sum));
}
    生成包含伪头的累加和(源，目的，长度，协议号)
static inline __wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr, unsigned short len, unsigned short proto, __wsum sum)
{
    __asm__(
            "addl %1, %0    ;\n"          //addl 加法
            "adcl %2, %0    ;\n"          //adcl 带进位的加法
            "adcl %3, %0    ;\n"
            "adcl $0, %0    ;\n"          //如果有进位，进行累加
            : "=r" (sum)
            : "g" (daddr), "g"(saddr), "g"((len + proto) << 8), "0"(sum)
           );

    return sum;
}
    将32位累加和折叠成16位校验和
static inline __sum16 csum_fold(__wsum sum)
{
    __asm__(
            "addl %1, %0            ;\n"
            "adcl $0xffff, %0       ;\n"
            : "=r" (sum)
            : "r" ((__force u32)sum << 16), "0" ((__force u32)sum & 0xffff0000)
           );
    return (__force __sum16)(~(__force u32)sum >> 16);
}
    基于伪头累加和，完成全包校验
static __inline__ int tcp_checksum_complete(struct sk_buff *skb)
{
    return skb->ip_summed != CHECKSUM_UNNECESSARY && __tcp_checksum_complete(skb);
}
__sum16 __skb_checksum_complete(struct sk_buff *skb)
{
    return __skb_checksum_complete_head(skb, skb->len);
}
__sum16 __skb_checksum_complete_head(struct sk_buff *skb, int len)
{
    __sum16 sum;

    sum = csum_fold(skb_checksum(skb, 0, len, skb->csum));
    if (likely(!sum)) {
        if (unlikely(skb->ip_summed == CHECKSUM_COMPLETE))
            netdev_rx_csum_fault(skb->dev);
        skb->ip_summed = CHECKSUM_UNNECESSARY;
    }
    return sum;
}
__wsum skb_checksum(const struct sk_buff *skb, int offset, int len, __wsum csum)
{
    int start = skb_headlen(skb);
    int i, copy = start - offset;
    int pos = 0;

    /* Checksum header. */
    if (copy > 0) {
        if (copy > len)
            copy = len;

        csum = csum_partial(skb->data + offset, copy, csum);
        if ((len -= copy) == 0)
            return csum;

        offset += copy;
        pos     = copy;
    }
    ......
}
    计算32位中间累加和
unsigned int csum_partial(const unsigned char * buff, int len, unsigned int sum)
{
    //arch/x86/lib/checksum_32.S 汇编文件
}
    基于TCP用户数据的中间累加和, 生成TCP包校验码
void tcp_v4_send_check(struct sock *sk, int len, struct sk_buff *skb)
{
    struct inet_sock *inet = inet_sk(sk);
    struct tcphdr *th = tcp_hdr(skb);

    if (skb->ip_summed == CHECKSUM_PARTIAL) {
        th->check = ~tcp_v4_check(len, inet->saddr, inet->daddr, 0); //附加伪头进行校验
        skb->csum_start = skb_transport_header(skb) - skb->head;
        skb->csum_offset = offsetof(struct tcphdr, check);
    } else {
        //完整的tcp校验和计算方法
        th->check = tcp_v4_check(len, inet->saddr, inet->daddr, csum_partial((char *)th, th->doff << 2, skb->csum));
    }
}
    在拷贝用户数据时同时计算累加和
unsigned int csum_partial_copy_nocheck(const char *src, char *dst, int len, int sum)
{
    return csum_partial_copy_generic(src, dst, len, sum, NULL, NULL); // arch/x86/lib/checksum_32.S
}
    ip头校验和计算
static inline __sum16 ip_fast_csum(const void *iph, unsigned int ihl)
{
    unsigned int sum;

    __asm__ __volatile__(
            "movl (%1), %0      ;\n"
            "subl $4, %2         ;\n"
            "jbe 2f              ;\n"
            "addl 4(%1), %0     ;\n"  //sum = sum + *(iph+4)
            "adcl 8(%1), %0     ;\n"  //sum = sum + *(iph+8) + carry
            "adcl 12(%1), %0    ;\n"  //sum = sum + *(iph+12) + carry
            "1:                adcl 16(%1), %0     ;\n" //sum = sum + *(iph+16) + carry
            "lea 4(%1), %1      ;\n"  //iph = iph + 4
            "decl %2            ;\n"
            "jne 1b             ;\n"
            "adcl $0, %0        ;\n"
            "movl %0, %2       ;\n"
            "shrl $16, %0       ;\n"
            "addw %w2, %w0   ;\n"
            "adcl $0, %0        ;\n"
            "notl %0            ;\n"
            "2:                                    ;\n"
            /* Since the input registers which are loaded with iph and ihl are modified, we must also specify them as outputs,
               or gcc will assume they contain their original values. */
            : "=r" (sum), "=r" (iph), "=r" (ihl)
            : "1" (iph), "2" (ihl)
                   : "memory"
                       );
    return (__force __sum16)sum;
}
    递减ip->ttl，更新校验和
static inline int ip_decrease_ttl(struct iphdr *iph)
{
    u32 check = (__force u32)iph->check;
    check += (__force u32)htons(0x0100);
    iph->check = (__force __sum16)(check + (check>=0xFFFF));
    return --iph->ttl;
}
static inline __wsum csum_add(__wsum csum, __wsum addend)
{
    u32 res = (__force u32)csum;
    res += (__force u32)addend;
    return (__force __wsum)(res + (res < (__force u32)addend));
}

static inline __wsum csum_sub(__wsum csum, __wsum addend)
{
    return csum_add(csum, ~addend);
}

static inline __wsum csum_block_add(__wsum csum, __wsum csum2, int offset)
{
    u32 sum = (__force u32)csum2;
    if (offset & 1)
        sum = ((sum & 0xFF00FF)<<8) + ((sum>>8) & 0xFF00FF);
    return csum_add(csum, (__force __wsum)sum);
}
static inline __wsum csum_block_sub(__wsum csum, __wsum csum2, int offset)
{
    u32 sum = (__force u32)csum2;
    if (offset & 1)
        sum = ((sum & 0xFF00FF)<<8) + ((sum>>8) & 0xFF00FF);
    return csum_sub(csum, (__force __wsum)sum);
}
[/函数实现]