/* Remove an entry from the "incomplete datagrams" queue, either
* because we completed, reassembled and processed it, or because
* it timed out.
*
* This is called _only_ from BH contexts, on packet reception
* processing and from frag queue expiration timers. -DaveM
*/
static void ip_free(struct ipq *qp)
{
struct ipfrag *fp;
/* Stop the timer for this entry. */
del_timer(&qp->timer);
/* Remove this entry from the "incomplete datagrams" queue. */
if(qp->next)
qp->next->pprev = qp->pprev;
*qp->pprev = qp->next;
/* Release all fragment data. */
fp = qp->fragments;
while (fp) {
struct ipfrag *xp = fp->next;
frag_kfree_skb(fp->skb);
frag_kfree_s(fp, sizeof(struct ipfrag));
fp = xp;
}
/* Release the IP header. */
frag_kfree_s(qp->iph, 64 + 8);
/* Finally, release the queue descriptor itself. */
frag_kfree_s(qp, sizeof(struct ipq));
}
// 从当前host中删除掉一个给定的ip报文队列
static void
ip_free(struct ipq * qp)
{
struct ipfrag *fp;
struct ipfrag *xp;
/* Stop the timer for this entry. */
del_timer(&qp->timer);
/* Remove this entry from the "incomplete datagrams" queue. */
// 如果队列是host中的第一个队列
if (qp->prev == NULL)
{
this_host->ipqueue = qp->next;
// 如果后面还有，则修改后面一个为第一个
if (this_host->ipqueue != NULL)
this_host->ipqueue->prev = NULL;
// 否则该删除的队列是最后一个，说明host已经没有队列了，删除host
else
rmthis_host();
}
// 否则不是第一个
else
{
qp->prev->next = qp->next;
// 如果不是最后一个，修改后一个的指针
if (qp->next != NULL)
qp->next->prev = qp->prev;
}
/* Release all fragment data. */
// 将这一个队列中所有的fragments全部释放掉
fp = qp->fragments;
while (fp != NULL)
{
xp = fp->next;
// 释放ipfrag中的ksb
frag_kfree_skb(fp->skb, FREE_READ);
// 再释放ipfrag结构
frag_kfree_s(fp, sizeof(struct ipfrag));
fp = xp;
}
/* Release the IP header. */
frag_kfree_s(qp->iph, 64 + 8);
/* Finally, release the queue descriptor itself. */
frag_kfree_s(qp, sizeof(struct ipq));
}
// 疑问: 这一个应该要修改this_host的 ip_frag_mem变量吧????
// 输入:
// ip头
// 返回:
// ip队列，这个ip队列已经被挂载到host上了
static struct ipq *
ip_create(struct ip * iph)
{
struct ipq *qp;
int ihlen;
// 调用malloc,申请一个空间
qp = (struct ipq *) frag_kmalloc(sizeof(struct ipq), GFP_ATOMIC);
if (qp == NULL)
{
// NETDEBUG(printk("IP: create: no memory left !\n"));
nids_params.no_mem("ip_create");
return (NULL);
}
// 填充为0
memset(qp, 0, sizeof(struct ipq));
/* Allocate memory for the IP header (plus 8 octets for ICMP). */
// 多分配8字节，需要保存ip头以及头后面8字节，因为icmp的内容就是ip头+8字节
ihlen = iph->ip_hl * 4;
qp->iph = (struct ip *) frag_kmalloc(64 + 8, GFP_ATOMIC);
if (qp->iph == NULL)
{
//NETDEBUG(printk("IP: create: no memory left !\n"));
nids_params.no_mem("ip_create");
frag_kfree_s(qp, sizeof(struct ipq));
return (NULL);
}
// 将ip头+8字节的内容保存在iph变量中。
// 到时候iph变量就能够直接作为icmp的内容，如果有必要发送icmp的话
memcpy(qp->iph, iph, ihlen + 8);
// 队列长度=0， 头长度，碎片队列
qp->len = 0;
qp->ihlen = ihlen;
qp->fragments = NULL;
// 挂载到当前host上
qp->hf = this_host;
/* Start a timer for this entry. */
// jiffies函数返回当前时间，毫秒为单位
qp->timer.expires = jiffies() + IP_FRAG_TIME; /* about 30 seconds */
// 设置关联
qp->timer.data = (unsigned long) qp; /* pointer to queue */
// 注册一个超时函数，当超时的时候，会回调这个函数
qp->timer.function = ip_expire; /* expire function */
// 将这个timer挂载到queue上
add_timer(&qp->timer);
/* Add this entry to the queue. */
// 将这个队列挂载到当前host上
qp->prev = NULL;
qp->next = this_host->ipqueue;
if (qp->next != NULL)
qp->next->prev = qp;
this_host->ipqueue = qp;
return (qp);
}
/*
Remove an entry from the "incomplete datagrams" queue, either
because we completed, reassembled and processed it, or because it
timed out.
*/
static void
//! ip_free(struct ipq * qp)
ip_free(struct ipq * qp,IP_THREAD_LOCAL_P ip_thread_local_p)
{
struct ipfrag *fp;
struct ipfrag *xp;
/* Stop the timer for this entry. */
//! del_timer(&qp->timer);
del_timer(&qp->timer,ip_thread_local_p);
/* Remove this entry from the "incomplete datagrams" queue. */
if (qp->prev == NULL) {
//! this_host->ipqueue = qp->next;
ip_thread_local_p->this_host->ipqueue = qp->next;
//! if (this_host->ipqueue != NULL)
if (ip_thread_local_p->this_host->ipqueue != NULL)
//! this_host->ipqueue->prev = NULL;
ip_thread_local_p->this_host->ipqueue->prev = NULL;
else
//! rmthis_host();
rmthis_host(ip_thread_local_p);
} else {
qp->prev->next = qp->next;
if (qp->next != NULL)
qp->next->prev = qp->prev;
}
/* Release all fragment data. */
fp = qp->fragments;
while (fp != NULL) {
xp = fp->next;
//! frag_kfree_skb(fp->skb, FREE_READ);
frag_kfree_skb(fp->skb, FREE_READ,ip_thread_local_p);
//! frag_kfree_s(fp, sizeof(struct ipfrag));
frag_kfree_s(fp, sizeof(struct ipfrag),ip_thread_local_p);
fp = xp;
}
/* Release the IP header. */
frag_kfree_s(qp->iph, 64 + 8, ip_thread_local_p);
/* Finally, release the queue descriptor itself. */
//! frag_kfree_s(qp, sizeof(struct ipq));
frag_kfree_s(qp, sizeof(struct ipq),ip_thread_local_p);
}
static void ip_free(struct ipq *qp)
{
struct ipfrag *fp;
struct ipfrag *xp;
/*
* Stop the timer for this entry.
*/
del_timer(&qp->timer);
/* Remove this entry from the "incomplete datagrams" queue. */
cli();
if (qp->prev == NULL)
{
ipqueue = qp->next;
if (ipqueue != NULL)
ipqueue->prev = NULL;
}
else
{
qp->prev->next = qp->next;
if (qp->next != NULL)
qp->next->prev = qp->prev;
}
/* Release all fragment data. */
fp = qp->fragments;
while (fp != NULL)
{
xp = fp->next;
IS_SKB(fp->skb);
frag_kfree_skb(fp->skb,FREE_READ);
frag_kfree_s(fp, sizeof(struct ipfrag));
fp = xp;
}
/* Release the IP header. */
frag_kfree_s(qp->iph, 64 + 8);
/* Finally, release the queue descriptor itself. */
frag_kfree_s(qp, sizeof(struct ipq));
sti();
}
static void fq_free(struct frag_queue *fq)
{
struct ipv6_frag *fp, *back;
del_timer(&fq->timer);
for (fp = fq->fragments; fp; ) {
frag_kfree_skb(fp->skb);
back = fp;
fp=fp->next;
frag_kfree_s(back, sizeof(*back));
}
fq->prev->next = fq->next;
fq->next->prev = fq->prev;
fq->prev = fq->next = NULL;
frag_kfree_s(fq, sizeof(*fq));
}
/*
Add an entry to the 'ipq' queue for a newly received IP datagram.
We will (hopefully :-) receive all other fragments of this datagram
in time, so we just create a queue for this datagram, in which we
will insert the received fragments at their respective positions.
*/
static struct ipq *
//! ip_create(struct ip * iph)
ip_create(struct ip * iph,IP_THREAD_LOCAL_P ip_thread_local_p) {
struct ipq *qp;
int ihlen;
//! qp = (struct ipq *) frag_kmalloc(sizeof(struct ipq), GFP_ATOMIC);
qp = (struct ipq *) frag_kmalloc(sizeof(struct ipq), GFP_ATOMIC,ip_thread_local_p);
if (qp == NULL) {
// NETDEBUG(printk("IP: create: no memory left !\n"));
nids_params.no_mem("ip_create");
return (NULL);
}
memset(qp, 0, sizeof(struct ipq));
/* Allocate memory for the IP header (plus 8 octets for ICMP). */
ihlen = iph->ip_hl * 4;
qp->iph = (struct ip *) frag_kmalloc(64 + 8, GFP_ATOMIC, ip_thread_local_p);
if (qp->iph == NULL) {
//NETDEBUG(printk("IP: create: no memory left !\n"));
nids_params.no_mem("ip_create");
//! frag_kfree_s(qp, sizeof(struct ipq));
frag_kfree_s(qp, sizeof(struct ipq),ip_thread_local_p);
return (NULL);
}
memcpy(qp->iph, iph, ihlen + 8);
qp->len = 0;
qp->ihlen = ihlen;
qp->fragments = NULL;
//! qp->hf = this_host;
qp->hf = ip_thread_local_p->this_host;
/* Start a timer for this entry. */
//! qp->timer.expires = jiffies() + IP_FRAG_TIME; /* about 30 seconds */
qp->timer.expires = jiffies(ip_thread_local_p) + IP_FRAG_TIME; /* about 30 seconds */
qp->timer.data = (unsigned long) qp; /* pointer to queue */
qp->timer.function = ip_expire; /* expire function */
//! add_timer(&qp->timer);
add_timer(&qp->timer,ip_thread_local_p);
/* Add this entry to the queue. */
qp->prev = NULL;
//! qp->next = this_host->ipqueue;
qp->next = ip_thread_local_p->this_host->ipqueue;
if (qp->next != NULL)
qp->next->prev = qp;
//! this_host->ipqueue = qp;
ip_thread_local_p->this_host->ipqueue = qp;
return (qp);
}
static struct ipq *ip_create(struct sk_buff *skb, struct iphdr *iph, struct device *dev)
{
struct ipq *qp;
int ihlen;
qp = (struct ipq *) frag_kmalloc(sizeof(struct ipq), GFP_ATOMIC);
if (qp == NULL)
{
NETDEBUG(printk("IP: create: no memory left !\n"));
return(NULL);
}
memset(qp, 0, sizeof(struct ipq));
/*
* Allocate memory for the IP header (plus 8 octets for ICMP).
*/
ihlen = iph->ihl * 4;
qp->iph = (struct iphdr *) frag_kmalloc(64 + 8, GFP_ATOMIC);
if (qp->iph == NULL)
{
NETDEBUG(printk("IP: create: no memory left !\n"));
frag_kfree_s(qp, sizeof(struct ipq));
return(NULL);
}
memcpy(qp->iph, iph, ihlen + 8);
qp->len = 0;
qp->ihlen = ihlen;
qp->fragments = NULL;
qp->dev = dev;
/* Start a timer for this entry. */
qp->timer.expires = jiffies + IP_FRAG_TIME; /* about 30 seconds */
qp->timer.data = (unsigned long) qp; /* pointer to queue */
qp->timer.function = ip_expire; /* expire function */
add_timer(&qp->timer);
/* Add this entry to the queue. */
qp->prev = NULL;
cli();
qp->next = ipqueue;
if (qp->next != NULL)
qp->next->prev = qp;
ipqueue = qp;
sti();
return(qp);
}
/* Add an entry to the 'ipq' queue for a newly received IP datagram.
* We will (hopefully :-) receive all other fragments of this datagram
* in time, so we just create a queue for this datagram, in which we
* will insert the received fragments at their respective positions.
*/
static struct ipq *ip_create(struct sk_buff *skb, struct iphdr *iph)
{
struct ipq *qp;
unsigned int hash;
int ihlen;
qp = (struct ipq *) frag_kmalloc(sizeof(struct ipq), GFP_ATOMIC);
if (qp == NULL)
goto out_nomem;
/* Allocate memory for the IP header (plus 8 octets for ICMP). */
ihlen = iph->ihl * 4;
qp->iph = (struct iphdr *) frag_kmalloc(64 + 8, GFP_ATOMIC);
if (qp->iph == NULL)
goto out_free;
memcpy(qp->iph, iph, ihlen + 8);
qp->len = 0;
qp->ihlen = ihlen;
qp->fragments = NULL;
qp->dev = skb->dev;
/* Initialize a timer for this entry. */
init_timer(&qp->timer);
qp->timer.expires = 0; /* (to be set later) */
qp->timer.data = (unsigned long) qp; /* pointer to queue */
qp->timer.function = ip_expire; /* expire function */
/* Add this entry to the queue. */
hash = ipqhashfn(iph->id, iph->saddr, iph->daddr, iph->protocol);
/* We are in a BH context, no locking necessary. -DaveM */
if((qp->next = ipq_hash[hash]) != NULL)
qp->next->pprev = &qp->next;
ipq_hash[hash] = qp;
qp->pprev = &ipq_hash[hash];
return qp;
out_free:
frag_kfree_s(qp, sizeof(struct ipq));
out_nomem:
NETDEBUG(printk(KERN_ERR "IP: create: no memory left !\n"));
return(NULL);
}
static void create_frag_entry(struct sk_buff *skb,
__u8 *nhptr,
struct frag_hdr *fhdr)
{
struct frag_queue *fq;
struct ipv6hdr *hdr;
fq = (struct frag_queue *) frag_kmalloc(sizeof(struct frag_queue),
GFP_ATOMIC);
if (fq == NULL) {
ipv6_statistics.Ip6ReasmFails++;
kfree_skb(skb);
return;
}
memset(fq, 0, sizeof(struct frag_queue));
fq->id = fhdr->identification;
hdr = skb->nh.ipv6h;
ipv6_addr_copy(&fq->saddr, &hdr->saddr);
ipv6_addr_copy(&fq->daddr, &hdr->daddr);
/* init_timer has been done by the memset */
fq->timer.function = frag_expire;
fq->timer.data = (long) fq;
fq->timer.expires = jiffies + sysctl_ip6frag_time;
reasm_queue(fq, skb, fhdr, nhptr);
if (fq->fragments) {
fq->prev = ipv6_frag_queue.prev;
fq->next = &ipv6_frag_queue;
fq->prev->next = fq;
ipv6_frag_queue.prev = fq;
add_timer(&fq->timer);
} else
frag_kfree_s(fq, sizeof(*fq));
}
/* Process an incoming IP datagram fragment. */
static char *
ip_defrag(struct ip *iph, struct sk_buff *skb)
{
struct ipfrag *prev, *next, *tmp;
struct ipfrag *tfp;
struct ipq *qp;
char *skb2;
unsigned char *ptr;
int flags, offset;
int i, ihl, end;
if (!hostfrag_find(iph) && skb)
hostfrag_create(iph);
/* Start by cleaning up the memory. */
if (this_host)
if (this_host->ip_frag_mem > IPFRAG_HIGH_THRESH)
ip_evictor();
/* Find the entry of this IP datagram in the "incomplete datagrams" queue. */
if (this_host)
qp = ip_find(iph);
else
qp = 0;
/* Is this a non-fragmented datagram? */
offset = ntohs(iph->ip_off);
flags = offset & ~IP_OFFSET;
offset &= IP_OFFSET;
if (((flags & IP_MF) == 0) && (offset == 0)) {
if (qp != NULL)
ip_free(qp); /* Fragmented frame replaced by full
unfragmented copy */
return 0;
}
/* ip_evictor() could have removed all queues for the current host */
if (!this_host)
hostfrag_create(iph);
offset <<= 3; /* offset is in 8-byte chunks */
ihl = iph->ip_hl * 4;
/*
If the queue already existed, keep restarting its timer as long as
we still are receiving fragments. Otherwise, create a fresh queue
entry.
*/
if (qp != NULL) {
/* ANK. If the first fragment is received, we should remember the correct
IP header (with options) */
if (offset == 0) {
qp->ihlen = ihl;
memcpy(qp->iph, iph, ihl + 8);
}
del_timer(&qp->timer);
qp->timer.expires = jiffies() + IP_FRAG_TIME; /* about 30 seconds */
qp->timer.data = (unsigned long) qp; /* pointer to queue */
qp->timer.function = ip_expire; /* expire function */
add_timer(&qp->timer);
}
else {
/* If we failed to create it, then discard the frame. */
if ((qp = ip_create(iph)) == NULL) {
kfree_skb(skb, FREE_READ);
return NULL;
}
}
/* Attempt to construct an oversize packet. */
if (ntohs(iph->ip_len) + (int) offset > 65535) {
// NETDEBUG(printk("Oversized packet received from %s\n", int_ntoa(iph->ip_src.s_addr)));
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERSIZED, iph, 0);
kfree_skb(skb, FREE_READ);
return NULL;
}
/* Determine the position of this fragment. */
end = offset + ntohs(iph->ip_len) - ihl;
/* Point into the IP datagram 'data' part. */
ptr = (unsigned char *)(skb->data + ihl);
/* Is this the final fragment? */
if ((flags & IP_MF) == 0)
qp->len = end;
/*
Find out which fragments are in front and at the back of us in the
chain of fragments so far. We must know where to put this
fragment, right?
*/
prev = NULL;
for (next = qp->fragments; next != NULL; next = next->next) {
if (next->offset >= offset)
break; /* bingo! */
prev = next;
}
/*
We found where to put this one. Check for overlap with preceding
fragment, and, if needed, align things so that any overlaps are
eliminated.
*/
if (prev != NULL && offset < prev->end) {
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERLAP, iph, 0);
i = prev->end - offset;
offset += i; /* ptr into datagram */
ptr += i; /* ptr into fragment data */
}
/*
Look for overlap with succeeding segments.
If we can merge fragments, do it.
*/
for (tmp = next; tmp != NULL; tmp = tfp) {
tfp = tmp->next;
if (tmp->offset >= end)
break; /* no overlaps at all */
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERLAP, iph, 0);
i = end - next->offset; /* overlap is 'i' bytes */
tmp->len -= i; /* so reduce size of */
tmp->offset += i; /* next fragment */
tmp->ptr += i;
/*
If we get a frag size of <= 0, remove it and the packet that it
goes with. We never throw the new frag away, so the frag being
dumped has always been charged for.
*/
if (tmp->len <= 0) {
if (tmp->prev != NULL)
tmp->prev->next = tmp->next;
else
qp->fragments = tmp->next;
if (tmp->next != NULL)
tmp->next->prev = tmp->prev;
next = tfp; /* We have killed the original next frame */
frag_kfree_skb(tmp->skb, FREE_READ);
frag_kfree_s(tmp, sizeof(struct ipfrag));
}
}
/* Insert this fragment in the chain of fragments. */
tfp = NULL;
tfp = ip_frag_create(offset, end, skb, ptr);
/*
No memory to save the fragment - so throw the lot. If we failed
the frag_create we haven't charged the queue.
*/
if (!tfp) {
nids_params.no_mem("ip_defrag");
kfree_skb(skb, FREE_READ);
return NULL;
}
/* From now on our buffer is charged to the queues. */
tfp->prev = prev;
tfp->next = next;
if (prev != NULL)
prev->next = tfp;
else
qp->fragments = tfp;
if (next != NULL)
next->prev = tfp;
/*
OK, so we inserted this new fragment into the chain. Check if we
now have a full IP datagram which we can bump up to the IP
layer...
*/
if (ip_done(qp)) {
skb2 = ip_glue(qp); /* glue together the fragments */
fprintf(stderr,"this is ipfragment.sd");
return (skb2);
}
return (NULL);
}
/* Process an incoming IP datagram fragment. */
struct sk_buff *ip_defrag(struct sk_buff *skb)
{
struct iphdr *iph = skb->nh.iph;
struct ipfrag *prev, *next, *tmp, *tfp;
struct ipq *qp;
unsigned char *ptr;
int flags, offset;
int i, ihl, end;
ip_statistics.IpReasmReqds++;
/* Start by cleaning up the memory. */
if (atomic_read(&ip_frag_mem) > sysctl_ipfrag_high_thresh)
ip_evictor();
/*
* Look for the entry for this IP datagram in the
* "incomplete datagrams" queue. If found, the
* timer is removed.
*/
qp = ip_find(iph, skb->dst);
/* Is this a non-fragmented datagram? */
offset = ntohs(iph->frag_off);
flags = offset & ~IP_OFFSET;
offset &= IP_OFFSET;
offset <<= 3; /* offset is in 8-byte chunks */
ihl = iph->ihl * 4;
/*
* Check whether to create a fresh queue entry. If the
* queue already exists, its timer will be restarted as
* long as we continue to receive fragments.
*/
if (qp) {
/* ANK. If the first fragment is received,
* we should remember the correct IP header (with options)
*/
if (offset == 0) {
/* Fragmented frame replaced by unfragmented copy? */
if ((flags & IP_MF) == 0)
goto out_freequeue;
qp->ihlen = ihl;
memcpy(qp->iph, iph, (ihl + 8));
}
} else {
/* Fragmented frame replaced by unfragmented copy? */
if ((offset == 0) && ((flags & IP_MF) == 0))
goto out_skb;
/* If we failed to create it, then discard the frame. */
qp = ip_create(skb, iph);
if (!qp)
goto out_freeskb;
}
/* Attempt to construct an oversize packet. */
if((ntohs(iph->tot_len) + ((int) offset)) > 65535)
goto out_oversize;
/* Determine the position of this fragment. */
end = offset + ntohs(iph->tot_len) - ihl;
/* Is this the final fragment? */
if ((flags & IP_MF) == 0)
qp->len = end;
/* Find out which fragments are in front and at the back of us
* in the chain of fragments so far. We must know where to put
* this fragment, right?
*/
prev = NULL;
for(next = qp->fragments; next != NULL; next = next->next) {
if (next->offset >= offset)
break; /* bingo! */
prev = next;
}
/* Point into the IP datagram 'data' part. */
ptr = skb->data + ihl;
/* We found where to put this one. Check for overlap with
* preceding fragment, and, if needed, align things so that
* any overlaps are eliminated.
*/
if ((prev != NULL) && (offset < prev->end)) {
i = prev->end - offset;
offset += i; /* ptr into datagram */
ptr += i; /* ptr into fragment data */
}
/* Look for overlap with succeeding segments.
* If we can merge fragments, do it.
*/
for (tmp = next; tmp != NULL; tmp = tfp) {
tfp = tmp->next;
if (tmp->offset >= end)
break; /* no overlaps at all */
i = end - next->offset; /* overlap is 'i' bytes */
tmp->len -= i; /* so reduce size of */
tmp->offset += i; /* next fragment */
tmp->ptr += i;
/* If we get a frag size of <= 0, remove it and the packet
* that it goes with.
*/
if (tmp->len <= 0) {
if (tmp->prev != NULL)
tmp->prev->next = tmp->next;
else
qp->fragments = tmp->next;
if (tmp->next != NULL)
tmp->next->prev = tmp->prev;
/* We have killed the original next frame. */
next = tfp;
frag_kfree_skb(tmp->skb);
frag_kfree_s(tmp, sizeof(struct ipfrag));
}
}
/*
* Create a fragment to hold this skb.
* No memory to save the fragment? throw the lot ...
*/
tfp = ip_frag_create(offset, end, skb, ptr);
if (!tfp)
goto out_freeskb;
/* Insert this fragment in the chain of fragments. */
tfp->prev = prev;
tfp->next = next;
if (prev != NULL)
prev->next = tfp;
else
qp->fragments = tfp;
if (next != NULL)
next->prev = tfp;
/* OK, so we inserted this new fragment into the chain.
* Check if we now have a full IP datagram which we can
* bump up to the IP layer...
*/
if (ip_done(qp)) {
/* Glue together the fragments. */
skb = ip_glue(qp);
/* Free the queue entry. */
out_freequeue:
ip_free(qp);
out_skb:
return skb;
}
/*
* The queue is still active ... reset its timer.
*/
out_timer:
mod_timer(&qp->timer, jiffies + sysctl_ipfrag_time); /* ~ 30 seconds */
out:
return NULL;
/*
* Error exits ... we need to reset the timer if there's a queue.
*/
out_oversize:
if (net_ratelimit())
printk(KERN_INFO "Oversized packet received from %d.%d.%d.%d\n",
NIPQUAD(iph->saddr));
/* the skb isn't in a fragment, so fall through to free it */
out_freeskb:
kfree_skb(skb);
ip_statistics.IpReasmFails++;
if (qp)
goto out_timer;
goto out;
}
// 每重组一个ip碎片，就会更新对应ipq的timer
//
//
static char *
ip_defrag(struct ip *iph, struct sk_buff *skb)
{
struct ipfrag *prev, *next, *tmp;
// 指向一个碎片
struct ipfrag *tfp;
// 指向一个队列
struct ipq *qp;
// 用来放返回值
char *skb2;
// 用来进行字节操作
unsigned char *ptr;
int flags, offset;
int i, ihl, end;
// 如果成功更新全局变量this_host, 并且skb是有内容的
if (!hostfrag_find(iph) && skb)
// 生成一个碎片
hostfrag_create(iph);
/* Start by cleaning up the memory. */
// 如果当前host不为空
if (this_host)
// 如果大于上限
if (this_host->ip_frag_mem > IPFRAG_HIGH_THRESH)
// 裁剪掉一些ip碎片，直到 ip_frag_mem < IPFRAG_LOW_THRESH
ip_evictor();
/* Find the entry of this IP datagram in the "incomplete datagrams" queue. */
// 如果host存在
if (this_host)
// 找到与这个ip头相关的ip队列
qp = ip_find(iph);
else
// 否则设置队列为空
qp = 0;
/* Is this a non-fragmented datagram? */
// ip_off是一个16位的字段，高3位用来保存标志信息，低12位用来保存当前碎片的偏移
offset = ntohs(iph->ip_off); /* 先把ip_offset这个字段取出来 */
// 把高3位取出来
flags = offset & ~IP_OFFSET;
// 把低13位取出来
offset &= IP_OFFSET;
// IP_MF==0表示当前收到是碎片后面没有碎片了，并且当前收到碎片是第一个碎片，
// 显然当前碎片虽在的ip报文仅仅有一个碎片
// 那么当前碎片(刚刚收到的碎片)并不需要重组，因此可以返回了
if (((flags & IP_MF) == 0) && (offset == 0))
{
// 如果队列不为空就释放掉
if (qp != NULL)
ip_free(qp); /* Fragmented frame replaced by full unfragmented copy */
return 0;
}
/* ip_evictor() could have removed all queues for the current host */
// 如果host全部被移出了，那么重新创建一个，针对当前ip头的host
// 但是，这个host并不会包括任何东西，它是一个空的，没有ip队列
if (!this_host)
hostfrag_create(iph);
// 计算offset和头长度
// 这个offset是刚刚收到的这个碎片的offset
offset <<= 3; /* offset is in 8-byte chunks */
ihl = iph->ip_hl * 4;
/*
If the queue already existed, keep restarting its timer as long as
we still are receiving fragments. Otherwise, create a fresh queue
entry.
*/
// 如果队列存在
if (qp != NULL)
{
/* ANK. If the first fragment is received, we should remember the correct
IP header (with options) */
// 如果偏移量为0，可能是该pi报文中的第一个碎片，因此要把前8字节保存下来
if (offset == 0)
{
// 保存头长度
qp->ihlen = ihl;
// 拷贝头信息+8字节
memcpy(qp->iph, iph, ihl + 8);
}
// 停止计时
del_timer(&qp->timer);
// 重新计时
qp->timer.expires = jiffies() + IP_FRAG_TIME; /* about 30 seconds */
// 设置关联
qp->timer.data = (unsigned long) qp; /* pointer to queue */
// 注册回调函数
qp->timer.function = ip_expire; /* expire function */
// 添加计数器
add_timer(&qp->timer);
}
// 否则队列不存在
else
{
/* If we failed to create it, then discard the frame. */
// 试图创建一个
if ((qp = ip_create(iph)) == NULL)
{
// 如果创建队列失败，那么释放当前碎片的空间并返回
kfree_skb(skb, FREE_READ);
return NULL;
}
}
/* Attempt to construct an oversize packet. */
// 如果头+长度 超长， 释放空间
if (ntohs(iph->ip_len) + (int) offset > 65535)
{
// NETDEBUG(printk("Oversized packet received from %s\n", int_ntoa(iph->ip_src.s_addr)));
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERSIZED, iph, 0);
kfree_skb(skb, FREE_READ);
return NULL;
}
/* Determine the position of this fragment. */
// 刚刚收到的碎片分组 + 刚刚收到的碎片ip包大小 - 刚刚收到的碎片ip包头大小
// = 刚刚收到的分组的结尾
end = offset + ntohs(iph->ip_len) - ihl;
/* Point into the IP datagram 'data' part. */
// 将指针，指向刚刚收到的碎片ip包的数据开头部分
// prt 指向肉
ptr = (unsigned char *)(skb->data + ihl);
/* Is this the final fragment? */
// 如果这是最后一个碎片，那么，整个队列的长度，就是当前的end
// 否则qp->len会在后面更改，因为会有重叠部分
if ((flags & IP_MF) == 0)
qp->len = end;
/*
Find out which fragments are in front and at the back of us in the
chain of fragments so far. We must know where to put this
fragment, right?
*/
prev = NULL;
// 给定一个offset,在所有的fragments中找到第一个
// 拥有不小于给定offset的offset的碎片，然后终止循环
// 由此可以猜测，这一个函数，其实是将某一个新来的碎片插入到
// 队列合适的位置，保证offset升序排列
// next 指向的是第一个不小于当前offset的碎片
// pre指向的是前一个碎片
for (next = qp->fragments; next != NULL; next = next->next)
{
if (next->offset >= offset)
break; /* bingo! */
prev = next;
}
/*--------------------------------------------------------------
注意:
next 指向的是第一个排在当前碎片后面的 碎片；
pre 指向的是第一个排在当前碎片前面的 碎片。
next 有可能offset与当前碎片的offset一样
----------------------------------------------------------------*/
/*
We found where to put this one. Check for overlap with preceding
fragment, and, if needed, align things so that any overlaps are
eliminated.
*/
// 如果有排在当前碎片前面的分组，并且， 该分组的结束比当前分组的offset大
// 说明有重叠，应噶修正当前offset，以先来的为准
if (prev != NULL && offset < prev->end)
{
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERLAP, iph, 0);
i = prev->end - offset;
// 将当前收到的碎片的offset增加i
offset += i; /* ptr into datagram */
// 将当前收到的碎片的指正向后移动i
ptr += i; /* ptr into fragment data */
}
/*
Look for overlap with succeeding segments.
If we can merge fragments, do it.
*/
// 现在往后查看是否有重叠的
// 从next开始
for (tmp = next; tmp != NULL; tmp = tfp)
{
// temp总是等于next
tfp = tmp->next;
// 如果next的 offset >= 当前分组的end，那就没有问题
if (tmp->offset >= end)
break; /* no overlaps at all */
// 否则警报
nids_params.syslog(NIDS_WARN_IP, NIDS_WARN_IP_OVERLAP, iph, 0);
// 记录当前的结束与下一个碎片的开始重叠多少
i = end - next->offset; /* overlap is 'i' bytes */
// 将next碎片的长度减少i
tmp->len -= i; /* so reduce size of */
// 将next碎片的开始增加i
tmp->offset += i; /* next fragment */
// 将指针也要向后移动i
tmp->ptr += i;
/*
If we get a frag size of <= 0, remove it and the packet that it
goes with. We never throw the new frag away, so the frag being
dumped has always been charged for.
*/
// 如果此时next碎片的长度小于0， 那么摘掉next节点
if (tmp->len <= 0)
{
if (tmp->prev != NULL)
tmp->prev->next = tmp->next;
else
qp->fragments = tmp->next;
if (tmp->next != NULL)
tmp->next->prev = tmp->prev;
// tfp原来就等于next->next，所以原来的next节点被摘掉了
next = tfp; /* We have killed the original "next" frame */
// 释放掉frag节点对应的内存
frag_kfree_skb(tmp->skb, FREE_READ);
// 释放掉frag节点
frag_kfree_s(tmp, sizeof(struct ipfrag));
}
}
/* Insert this fragment in the chain of fragments. */
tfp = NULL;
// offset是刚刚接收到的碎片偏移，按字节
// end 是刚刚接收到的碎片的结束字节
// skb 是传进来的一个内存空间，应该事先分配好
// prt 是指向这个碎片第一个数据的指针
tfp = ip_frag_create(offset, end, skb, ptr);
/*---------------------------------------------------
注意:
ip_frag_create函数只是创建一个ip_frag，还没有吧
它挂载到队列中
-----------------------------------------------------*/
/*
No memory to save the fragment - so throw the lot. If we failed
the frag_create we haven't charged the queue.
*/
if (!tfp)
{
nids_params.no_mem("ip_defrag");
kfree_skb(skb, FREE_READ);
return NULL;
}
/* From now on our buffer is charged to the queues. */
// 将刚刚创建的ip_frag挂载到队列中去了
tfp->prev = prev;
tfp->next = next;
if (prev != NULL)
prev->next = tfp;
else
qp->fragments = tfp;
if (next != NULL)
next->prev = tfp;
/*
OK, so we inserted this new fragment into the chain. Check if we
now have a full IP datagram which we can bump up to the IP
layer...
*/
// 检查是否完整
if (ip_done(qp))
{
skb2 = ip_glue(qp); /* glue together the fragments */
return (skb2);
}
// 如果没有完整，那么返回空，继续执行
return (NULL);
}
/*
* check if this fragment completes the packet
* returns true on success
*/
static u8* reasm_frag(struct frag_queue *fq, struct sk_buff **skb_in)
{
struct ipv6_frag *fp;
struct ipv6_frag *head = fq->fragments;
struct ipv6_frag *tail = NULL;
struct sk_buff *skb;
__u32 offset = 0;
__u32 payload_len;
__u16 unfrag_len;
__u16 copy;
u8 *nhptr;
for(fp = head; fp; fp=fp->next) {
if (offset != fp->offset)
return NULL;
offset += fp->len;
tail = fp;
}
/*
* we know the m_flag arrived and we have a queue,
* starting from 0, without gaps.
* this means we have all fragments.
*/
/* Unfragmented part is taken from the first segment.
(fixed --ANK (980728))
*/
unfrag_len = (u8 *) (head->fhdr) - (u8 *) (head->skb->nh.ipv6h + 1);
payload_len = (unfrag_len + tail->offset +
(tail->skb->tail - (__u8 *) (tail->fhdr + 1)));
if (payload_len > 65535) {
if (net_ratelimit())
printk(KERN_DEBUG "reasm_frag: payload len = %d\n", payload_len);
ipv6_statistics.Ip6ReasmFails++;
fq_free(fq);
return NULL;
}
if ((skb = dev_alloc_skb(sizeof(struct ipv6hdr) + payload_len))==NULL) {
if (net_ratelimit())
printk(KERN_DEBUG "reasm_frag: no memory for reassembly\n");
ipv6_statistics.Ip6ReasmFails++;
fq_free(fq);
return NULL;
}
copy = unfrag_len + sizeof(struct ipv6hdr);
skb->nh.ipv6h = (struct ipv6hdr *) skb->data;
skb->dev = fq->dev;
skb->protocol = __constant_htons(ETH_P_IPV6);
skb->pkt_type = head->skb->pkt_type;
memcpy(skb->cb, head->skb->cb, sizeof(skb->cb));
skb->dst = dst_clone(head->skb->dst);
memcpy(skb_put(skb, copy), head->skb->nh.ipv6h, copy);
nhptr = skb->nh.raw + fq->nhoffset;
*nhptr = fq->nexthdr;
skb->h.raw = skb->tail;
skb->nh.ipv6h->payload_len = ntohs(payload_len);
*skb_in = skb;
/*
* FIXME: If we don't have a checksum we ought to be able
* to defragment and checksum in this pass. [AC]
* Note that we don't really know yet whether the protocol
* needs checksums at all. It might still be a good idea. -AK
*/
for(fp = fq->fragments; fp; ) {
struct ipv6_frag *back;
memcpy(skb_put(skb, fp->len), (__u8*)(fp->fhdr + 1), fp->len);
frag_kfree_skb(fp->skb);
back = fp;
fp=fp->next;
frag_kfree_s(back, sizeof(*back));
}
del_timer(&fq->timer);
fq->prev->next = fq->next;
fq->next->prev = fq->prev;
fq->prev = fq->next = NULL;
frag_kfree_s(fq, sizeof(*fq));
ipv6_statistics.Ip6ReasmOKs++;
return nhptr;
}
static void reasm_queue(struct frag_queue *fq, struct sk_buff *skb,
struct frag_hdr *fhdr, u8 *nhptr)
{
struct ipv6_frag *nfp, *fp, **bptr;
nfp = (struct ipv6_frag *) frag_kmalloc(sizeof(struct ipv6_frag),
GFP_ATOMIC);
if (nfp == NULL) {
kfree_skb(skb);
return;
}
nfp->offset = ntohs(fhdr->frag_off) & ~0x7;
nfp->len = (ntohs(skb->nh.ipv6h->payload_len) -
((u8 *) (fhdr + 1) - (u8 *) (skb->nh.ipv6h + 1)));
if ((u32)nfp->offset + (u32)nfp->len >= 65536) {
icmpv6_param_prob(skb,ICMPV6_HDR_FIELD, (u8*)&fhdr->frag_off);
goto err;
}
if (fhdr->frag_off & __constant_htons(0x0001)) {
/* Check if the fragment is rounded to 8 bytes.
* Required by the RFC.
* ... and would break our defragmentation algorithm 8)
*/
if (nfp->len & 0x7) {
printk(KERN_DEBUG "fragment not rounded to 8bytes\n");
/*
It is not in specs, but I see no reasons
to send an error in this case. --ANK
*/
if (nfp->offset == 0)
icmpv6_param_prob(skb, ICMPV6_HDR_FIELD,
&skb->nh.ipv6h->payload_len);
goto err;
}
}
nfp->skb = skb;
nfp->fhdr = fhdr;
nfp->next = NULL;
bptr = &fq->fragments;
for (fp = fq->fragments; fp; fp=fp->next) {
if (nfp->offset <= fp->offset)
break;
bptr = &fp->next;
}
if (fp && fp->offset == nfp->offset) {
if (nfp->len != fp->len) {
printk(KERN_DEBUG "reasm_queue: dup with wrong len\n");
}
/* duplicate. discard it. */
goto err;
}
atomic_add(skb->truesize, &ip6_frag_mem);
/* All the checks are done, fragment is acepted.
Only now we are allowed to update reassembly data!
(fixed --ANK (980728))
*/
/* iif always set to one of the last arrived segment */
fq->dev = skb->dev;
fq->iif = skb->dev->ifindex;
/* Last fragment */
if ((fhdr->frag_off & __constant_htons(0x0001)) == 0)
fq->last_in |= LAST_IN;
/* First fragment.
nexthdr and nhptr are get from the first fragment.
Moreover, nexthdr is UNDEFINED for all the fragments but the
first one.
(fixed --ANK (980728))
*/
if (nfp->offset == 0) {
fq->nexthdr = fhdr->nexthdr;
fq->last_in |= FIRST_IN;
fq->nhoffset = nhptr - skb->nh.raw;
}
*bptr = nfp;
nfp->next = fp;
return;
err:
frag_kfree_s(nfp, sizeof(*nfp));
kfree_skb(skb);
}
