Commit d2c4a075 authored by David S. Miller's avatar David S. Miller

Merge http://kernel-acme.bkbits.net:8080/net-cleanups-2.5

into nuts.ninka.net:/home/davem/src/BK/net-2.5
parents 6445be80 49ab902f
...@@ -261,7 +261,7 @@ ...@@ -261,7 +261,7 @@
int sysctl_tcp_fin_timeout = TCP_FIN_TIMEOUT; int sysctl_tcp_fin_timeout = TCP_FIN_TIMEOUT;
struct tcp_mib tcp_statistics[NR_CPUS*2]; struct tcp_mib tcp_statistics[NR_CPUS * 2];
kmem_cache_t *tcp_openreq_cachep; kmem_cache_t *tcp_openreq_cachep;
kmem_cache_t *tcp_bucket_cachep; kmem_cache_t *tcp_bucket_cachep;
...@@ -270,8 +270,8 @@ kmem_cache_t *tcp_timewait_cachep; ...@@ -270,8 +270,8 @@ kmem_cache_t *tcp_timewait_cachep;
atomic_t tcp_orphan_count = ATOMIC_INIT(0); atomic_t tcp_orphan_count = ATOMIC_INIT(0);
int sysctl_tcp_mem[3]; int sysctl_tcp_mem[3];
int sysctl_tcp_wmem[3] = { 4*1024, 16*1024, 128*1024 }; int sysctl_tcp_wmem[3] = { 4 * 1024, 16 * 1024, 128 * 1024 };
int sysctl_tcp_rmem[3] = { 4*1024, 87380, 87380*2 }; int sysctl_tcp_rmem[3] = { 4 * 1024, 87380, 87380 * 2 };
atomic_t tcp_memory_allocated; /* Current allocated memory. */ atomic_t tcp_memory_allocated; /* Current allocated memory. */
atomic_t tcp_sockets_allocated; /* Current number of TCP sockets. */ atomic_t tcp_sockets_allocated; /* Current number of TCP sockets. */
...@@ -282,13 +282,13 @@ atomic_t tcp_sockets_allocated; /* Current number of TCP sockets. */ ...@@ -282,13 +282,13 @@ atomic_t tcp_sockets_allocated; /* Current number of TCP sockets. */
* is strict, actions are advisory and have some latency. */ * is strict, actions are advisory and have some latency. */
int tcp_memory_pressure; int tcp_memory_pressure;
#define TCP_PAGES(amt) (((amt)+TCP_MEM_QUANTUM-1)/TCP_MEM_QUANTUM) #define TCP_PAGES(amt) (((amt) + TCP_MEM_QUANTUM - 1) / TCP_MEM_QUANTUM)
int tcp_mem_schedule(struct sock *sk, int size, int kind) int tcp_mem_schedule(struct sock *sk, int size, int kind)
{ {
int amt = TCP_PAGES(size); int amt = TCP_PAGES(size);
sk->forward_alloc += amt*TCP_MEM_QUANTUM; sk->forward_alloc += amt * TCP_MEM_QUANTUM;
atomic_add(amt, &tcp_memory_allocated); atomic_add(amt, &tcp_memory_allocated);
/* Under limit. */ /* Under limit. */
...@@ -317,25 +317,26 @@ int tcp_mem_schedule(struct sock *sk, int size, int kind) ...@@ -317,25 +317,26 @@ int tcp_mem_schedule(struct sock *sk, int size, int kind)
} }
if (!tcp_memory_pressure || if (!tcp_memory_pressure ||
sysctl_tcp_mem[2] > atomic_read(&tcp_sockets_allocated) sysctl_tcp_mem[2] > atomic_read(&tcp_sockets_allocated) *
* TCP_PAGES(sk->wmem_queued+atomic_read(&sk->rmem_alloc)+ TCP_PAGES(sk->wmem_queued +
atomic_read(&sk->rmem_alloc) +
sk->forward_alloc)) sk->forward_alloc))
return 1; return 1;
suppress_allocation: suppress_allocation:
if (kind == 0) { if (!kind) {
tcp_moderate_sndbuf(sk); tcp_moderate_sndbuf(sk);
/* Fail only if socket is _under_ its sndbuf. /* Fail only if socket is _under_ its sndbuf.
* In this case we cannot block, so that we have to fail. * In this case we cannot block, so that we have to fail.
*/ */
if (sk->wmem_queued+size >= sk->sndbuf) if (sk->wmem_queued + size >= sk->sndbuf)
return 1; return 1;
} }
/* Alas. Undo changes. */ /* Alas. Undo changes. */
sk->forward_alloc -= amt*TCP_MEM_QUANTUM; sk->forward_alloc -= amt * TCP_MEM_QUANTUM;
atomic_sub(amt, &tcp_memory_allocated); atomic_sub(amt, &tcp_memory_allocated);
return 0; return 0;
} }
...@@ -343,8 +344,9 @@ int tcp_mem_schedule(struct sock *sk, int size, int kind) ...@@ -343,8 +344,9 @@ int tcp_mem_schedule(struct sock *sk, int size, int kind)
void __tcp_mem_reclaim(struct sock *sk) void __tcp_mem_reclaim(struct sock *sk)
{ {
if (sk->forward_alloc >= TCP_MEM_QUANTUM) { if (sk->forward_alloc >= TCP_MEM_QUANTUM) {
atomic_sub(sk->forward_alloc/TCP_MEM_QUANTUM, &tcp_memory_allocated); atomic_sub(sk->forward_alloc / TCP_MEM_QUANTUM,
sk->forward_alloc &= (TCP_MEM_QUANTUM-1); &tcp_memory_allocated);
sk->forward_alloc &= TCP_MEM_QUANTUM - 1;
if (tcp_memory_pressure && if (tcp_memory_pressure &&
atomic_read(&tcp_memory_allocated) < sysctl_tcp_mem[0]) atomic_read(&tcp_memory_allocated) < sysctl_tcp_mem[0])
tcp_memory_pressure = 0; tcp_memory_pressure = 0;
...@@ -362,7 +364,8 @@ void tcp_rfree(struct sk_buff *skb) ...@@ -362,7 +364,8 @@ void tcp_rfree(struct sk_buff *skb)
/* /*
* LISTEN is a special case for poll.. * LISTEN is a special case for poll..
*/ */
static __inline__ unsigned int tcp_listen_poll(struct sock *sk, poll_table *wait) static __inline__ unsigned int tcp_listen_poll(struct sock *sk,
poll_table *wait)
{ {
return tcp_sk(sk)->accept_queue ? (POLLIN | POLLRDNORM) : 0; return tcp_sk(sk)->accept_queue ? (POLLIN | POLLRDNORM) : 0;
} }
...@@ -374,7 +377,7 @@ static __inline__ unsigned int tcp_listen_poll(struct sock *sk, poll_table *wait ...@@ -374,7 +377,7 @@ static __inline__ unsigned int tcp_listen_poll(struct sock *sk, poll_table *wait
* take care of normal races (between the test and the event) and we don't * take care of normal races (between the test and the event) and we don't
* go look at any of the socket buffers directly. * go look at any of the socket buffers directly.
*/ */
unsigned int tcp_poll(struct file * file, struct socket *sock, poll_table *wait) unsigned int tcp_poll(struct file *file, struct socket *sock, poll_table *wait)
{ {
unsigned int mask; unsigned int mask;
struct sock *sk = sock->sk; struct sock *sk = sock->sk;
...@@ -426,13 +429,13 @@ unsigned int tcp_poll(struct file * file, struct socket *sock, poll_table *wait) ...@@ -426,13 +429,13 @@ unsigned int tcp_poll(struct file * file, struct socket *sock, poll_table *wait)
mask |= POLLIN | POLLRDNORM; mask |= POLLIN | POLLRDNORM;
/* Connected? */ /* Connected? */
if ((1 << sk->state) & ~(TCPF_SYN_SENT|TCPF_SYN_RECV)) { if ((1 << sk->state) & ~(TCPF_SYN_SENT | TCPF_SYN_RECV)) {
/* Potential race condition. If read of tp below will /* Potential race condition. If read of tp below will
* escape above sk->state, we can be illegally awaken * escape above sk->state, we can be illegally awaken
* in SYN_* states. */ * in SYN_* states. */
if ((tp->rcv_nxt != tp->copied_seq) && if ((tp->rcv_nxt != tp->copied_seq) &&
(tp->urg_seq != tp->copied_seq || (tp->urg_seq != tp->copied_seq ||
tp->rcv_nxt != tp->copied_seq+1 || tp->rcv_nxt != tp->copied_seq + 1 ||
sk->urginline || !tp->urg_data)) sk->urginline || !tp->urg_data))
mask |= POLLIN | POLLRDNORM; mask |= POLLIN | POLLRDNORM;
...@@ -471,7 +474,7 @@ void tcp_write_space(struct sock *sk) ...@@ -471,7 +474,7 @@ void tcp_write_space(struct sock *sk)
if (sk->sleep && waitqueue_active(sk->sleep)) if (sk->sleep && waitqueue_active(sk->sleep))
wake_up_interruptible(sk->sleep); wake_up_interruptible(sk->sleep);
if (sock->fasync_list && !(sk->shutdown&SEND_SHUTDOWN)) if (sock->fasync_list && !(sk->shutdown & SEND_SHUTDOWN))
sock_wake_async(sock, 2, POLL_OUT); sock_wake_async(sock, 2, POLL_OUT);
} }
} }
...@@ -481,42 +484,41 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) ...@@ -481,42 +484,41 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg)
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
int answ; int answ;
switch(cmd) { switch (cmd) {
case SIOCINQ: case SIOCINQ:
if (sk->state == TCP_LISTEN) if (sk->state == TCP_LISTEN)
return(-EINVAL); return -EINVAL;
lock_sock(sk); lock_sock(sk);
if ((1<<sk->state) & (TCPF_SYN_SENT|TCPF_SYN_RECV)) if ((1 << sk->state) & (TCPF_SYN_SENT | TCPF_SYN_RECV))
answ = 0; answ = 0;
else if (sk->urginline || !tp->urg_data || else if (sk->urginline || !tp->urg_data ||
before(tp->urg_seq,tp->copied_seq) || before(tp->urg_seq, tp->copied_seq) ||
!before(tp->urg_seq,tp->rcv_nxt)) { !before(tp->urg_seq, tp->rcv_nxt)) {
answ = tp->rcv_nxt - tp->copied_seq; answ = tp->rcv_nxt - tp->copied_seq;
/* Subtract 1, if FIN is in queue. */ /* Subtract 1, if FIN is in queue. */
if (answ && !skb_queue_empty(&sk->receive_queue)) if (answ && !skb_queue_empty(&sk->receive_queue))
answ -= ((struct sk_buff*)sk->receive_queue.prev)->h.th->fin; answ -=
((struct sk_buff*)sk->receive_queue.prev)->h.th->fin;
} else } else
answ = tp->urg_seq - tp->copied_seq; answ = tp->urg_seq - tp->copied_seq;
release_sock(sk); release_sock(sk);
break; break;
case SIOCATMARK: case SIOCATMARK:
{
answ = tp->urg_data && tp->urg_seq == tp->copied_seq; answ = tp->urg_data && tp->urg_seq == tp->copied_seq;
break; break;
}
case SIOCOUTQ: case SIOCOUTQ:
if (sk->state == TCP_LISTEN) if (sk->state == TCP_LISTEN)
return(-EINVAL); return -EINVAL;
if ((1<<sk->state) & (TCPF_SYN_SENT|TCPF_SYN_RECV)) if ((1 << sk->state) & (TCPF_SYN_SENT | TCPF_SYN_RECV))
answ = 0; answ = 0;
else else
answ = tp->write_seq - tp->snd_una; answ = tp->write_seq - tp->snd_una;
break; break;
default: default:
return(-ENOIOCTLCMD); return -ENOIOCTLCMD;
}; };
return put_user(answ, (int *)arg); return put_user(answ, (int *)arg);
...@@ -541,7 +543,7 @@ int tcp_listen_start(struct sock *sk) ...@@ -541,7 +543,7 @@ int tcp_listen_start(struct sock *sk)
memset(lopt, 0, sizeof(struct tcp_listen_opt)); memset(lopt, 0, sizeof(struct tcp_listen_opt));
for (lopt->max_qlen_log = 6; ; lopt->max_qlen_log++) for (lopt->max_qlen_log = 6; ; lopt->max_qlen_log++)
if ((1<<lopt->max_qlen_log) >= sysctl_max_syn_backlog) if ((1 << lopt->max_qlen_log) >= sysctl_max_syn_backlog)
break; break;
write_lock_bh(&tp->syn_wait_lock); write_lock_bh(&tp->syn_wait_lock);
...@@ -588,12 +590,12 @@ static void tcp_listen_stop (struct sock *sk) ...@@ -588,12 +590,12 @@ static void tcp_listen_stop (struct sock *sk)
/* make all the listen_opt local to us */ /* make all the listen_opt local to us */
write_lock_bh(&tp->syn_wait_lock); write_lock_bh(&tp->syn_wait_lock);
tp->listen_opt =NULL; tp->listen_opt = NULL;
write_unlock_bh(&tp->syn_wait_lock); write_unlock_bh(&tp->syn_wait_lock);
tp->accept_queue = tp->accept_queue_tail = NULL; tp->accept_queue = tp->accept_queue_tail = NULL;
if (lopt->qlen) { if (lopt->qlen) {
for (i=0; i<TCP_SYNQ_HSIZE; i++) { for (i = 0; i < TCP_SYNQ_HSIZE; i++) {
while ((req = lopt->syn_table[i]) != NULL) { while ((req = lopt->syn_table[i]) != NULL) {
lopt->syn_table[i] = req->dl_next; lopt->syn_table[i] = req->dl_next;
lopt->qlen--; lopt->qlen--;
...@@ -610,18 +612,18 @@ static void tcp_listen_stop (struct sock *sk) ...@@ -610,18 +612,18 @@ static void tcp_listen_stop (struct sock *sk)
} }
} }
} }
BUG_TRAP(lopt->qlen == 0); BUG_TRAP(!lopt->qlen);
kfree(lopt); kfree(lopt);
while ((req=acc_req) != NULL) { while ((req = acc_req) != NULL) {
struct sock *child = req->sk; struct sock *child = req->sk;
acc_req = req->dl_next; acc_req = req->dl_next;
local_bh_disable(); local_bh_disable();
bh_lock_sock(child); bh_lock_sock(child);
BUG_TRAP(child->lock.users==0); BUG_TRAP(!child->lock.users);
sock_hold(child); sock_hold(child);
tcp_disconnect(child, O_NONBLOCK); tcp_disconnect(child, O_NONBLOCK);
...@@ -639,7 +641,7 @@ static void tcp_listen_stop (struct sock *sk) ...@@ -639,7 +641,7 @@ static void tcp_listen_stop (struct sock *sk)
tcp_acceptq_removed(sk); tcp_acceptq_removed(sk);
tcp_openreq_fastfree(req); tcp_openreq_fastfree(req);
} }
BUG_TRAP(sk->ack_backlog == 0); BUG_TRAP(!sk->ack_backlog);
} }
/* /*
...@@ -647,21 +649,20 @@ static void tcp_listen_stop (struct sock *sk) ...@@ -647,21 +649,20 @@ static void tcp_listen_stop (struct sock *sk)
* *
* Note: Must be called with the socket locked. * Note: Must be called with the socket locked.
*/ */
static int wait_for_tcp_connect(struct sock * sk, int flags, long *timeo_p) static int wait_for_tcp_connect(struct sock *sk, int flags, long *timeo_p)
{ {
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
struct task_struct *tsk = current; struct task_struct *tsk = current;
DECLARE_WAITQUEUE(wait, tsk); DECLARE_WAITQUEUE(wait, tsk);
while((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) { while ((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) {
if(sk->err) if (sk->err)
return sock_error(sk); return sock_error(sk);
if((1 << sk->state) & if ((1 << sk->state) & ~(TCPF_SYN_SENT | TCPF_SYN_RECV))
~(TCPF_SYN_SENT | TCPF_SYN_RECV))
return -EPIPE; return -EPIPE;
if(!*timeo_p) if (!*timeo_p)
return -EAGAIN; return -EAGAIN;
if(signal_pending(tsk)) if (signal_pending(tsk))
return sock_intr_errno(*timeo_p); return sock_intr_errno(*timeo_p);
__set_task_state(tsk, TASK_INTERRUPTIBLE); __set_task_state(tsk, TASK_INTERRUPTIBLE);
...@@ -687,7 +688,7 @@ static inline int tcp_memory_free(struct sock *sk) ...@@ -687,7 +688,7 @@ static inline int tcp_memory_free(struct sock *sk)
/* /*
* Wait for more memory for a socket * Wait for more memory for a socket
*/ */
static int wait_for_tcp_memory(struct sock * sk, long *timeo) static int wait_for_tcp_memory(struct sock *sk, long *timeo)
{ {
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
int err = 0; int err = 0;
...@@ -696,7 +697,7 @@ static int wait_for_tcp_memory(struct sock * sk, long *timeo) ...@@ -696,7 +697,7 @@ static int wait_for_tcp_memory(struct sock * sk, long *timeo)
DECLARE_WAITQUEUE(wait, current); DECLARE_WAITQUEUE(wait, current);
if (tcp_memory_free(sk)) if (tcp_memory_free(sk))
current_timeo = vm_wait = (net_random()%(HZ/5))+2; current_timeo = vm_wait = (net_random() % (HZ / 5)) + 2;
add_wait_queue(sk->sleep, &wait); add_wait_queue(sk->sleep, &wait);
for (;;) { for (;;) {
...@@ -748,27 +749,28 @@ static int wait_for_tcp_memory(struct sock * sk, long *timeo) ...@@ -748,27 +749,28 @@ static int wait_for_tcp_memory(struct sock * sk, long *timeo)
goto out; goto out;
} }
ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags); ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset,
size_t psize, int flags);
static inline int static inline int can_coalesce(struct sk_buff *skb, int i, struct page *page,
can_coalesce(struct sk_buff *skb, int i, struct page *page, int off) int off)
{ {
if (i) { if (i) {
skb_frag_t *frag = &skb_shinfo(skb)->frags[i-1]; skb_frag_t *frag = &skb_shinfo(skb)->frags[i - 1];
return page == frag->page && return page == frag->page &&
off == frag->page_offset+frag->size; off == frag->page_offset + frag->size;
} }
return 0; return 0;
} }
static inline void static inline void fill_page_desc(struct sk_buff *skb, int i,
fill_page_desc(struct sk_buff *skb, int i, struct page *page, int off, int size) struct page *page, int off, int size)
{ {
skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
frag->page = page; frag->page = page;
frag->page_offset = off; frag->page_offset = off;
frag->size = size; frag->size = size;
skb_shinfo(skb)->nr_frags = i+1; skb_shinfo(skb)->nr_frags = i + 1;
} }
static inline void tcp_mark_push(struct tcp_opt *tp, struct sk_buff *skb) static inline void tcp_mark_push(struct tcp_opt *tp, struct sk_buff *skb)
...@@ -779,11 +781,11 @@ static inline void tcp_mark_push(struct tcp_opt *tp, struct sk_buff *skb) ...@@ -779,11 +781,11 @@ static inline void tcp_mark_push(struct tcp_opt *tp, struct sk_buff *skb)
static inline int forced_push(struct tcp_opt *tp) static inline int forced_push(struct tcp_opt *tp)
{ {
return after(tp->write_seq, tp->pushed_seq + (tp->max_window>>1)); return after(tp->write_seq, tp->pushed_seq + (tp->max_window >> 1));
} }
static inline void static inline void skb_entail(struct sock *sk, struct tcp_opt *tp,
skb_entail(struct sock *sk, struct tcp_opt *tp, struct sk_buff *skb) struct sk_buff *skb)
{ {
skb->csum = 0; skb->csum = 0;
TCP_SKB_CB(skb)->seq = tp->write_seq; TCP_SKB_CB(skb)->seq = tp->write_seq;
...@@ -792,12 +794,12 @@ skb_entail(struct sock *sk, struct tcp_opt *tp, struct sk_buff *skb) ...@@ -792,12 +794,12 @@ skb_entail(struct sock *sk, struct tcp_opt *tp, struct sk_buff *skb)
TCP_SKB_CB(skb)->sacked = 0; TCP_SKB_CB(skb)->sacked = 0;
__skb_queue_tail(&sk->write_queue, skb); __skb_queue_tail(&sk->write_queue, skb);
tcp_charge_skb(sk, skb); tcp_charge_skb(sk, skb);
if (tp->send_head == NULL) if (!tp->send_head)
tp->send_head = skb; tp->send_head = skb;
} }
static inline void static inline void tcp_mark_urg(struct tcp_opt *tp, int flags,
tcp_mark_urg(struct tcp_opt *tp, int flags, struct sk_buff *skb) struct sk_buff *skb)
{ {
if (flags & MSG_OOB) { if (flags & MSG_OOB) {
tp->urg_mode = 1; tp->urg_mode = 1;
...@@ -806,15 +808,16 @@ tcp_mark_urg(struct tcp_opt *tp, int flags, struct sk_buff *skb) ...@@ -806,15 +808,16 @@ tcp_mark_urg(struct tcp_opt *tp, int flags, struct sk_buff *skb)
} }
} }
static inline void static inline void tcp_push(struct sock *sk, struct tcp_opt *tp, int flags,
tcp_push(struct sock *sk, struct tcp_opt *tp, int flags, int mss_now, int nonagle) int mss_now, int nonagle)
{ {
if (tp->send_head) { if (tp->send_head) {
struct sk_buff *skb = sk->write_queue.prev; struct sk_buff *skb = sk->write_queue.prev;
if (!(flags&MSG_MORE) || forced_push(tp)) if (!(flags & MSG_MORE) || forced_push(tp))
tcp_mark_push(tp, skb); tcp_mark_push(tp, skb);
tcp_mark_urg(tp, flags, skb); tcp_mark_urg(tp, flags, skb);
__tcp_push_pending_frames(sk, tp, mss_now, (flags&MSG_MORE) ? 2 : nonagle); __tcp_push_pending_frames(sk, tp, mss_now,
(flags & MSG_MORE) ? 2 : nonagle);
} }
} }
...@@ -822,22 +825,23 @@ static int tcp_error(struct sock *sk, int flags, int err) ...@@ -822,22 +825,23 @@ static int tcp_error(struct sock *sk, int flags, int err)
{ {
if (err == -EPIPE) if (err == -EPIPE)
err = sock_error(sk) ? : -EPIPE; err = sock_error(sk) ? : -EPIPE;
if (err == -EPIPE && !(flags&MSG_NOSIGNAL)) if (err == -EPIPE && !(flags & MSG_NOSIGNAL))
send_sig(SIGPIPE, current, 0); send_sig(SIGPIPE, current, 0);
return err; return err;
} }
ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size_t psize, int flags) ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset,
size_t psize, int flags)
{ {
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
int mss_now; int mss_now;
int err; int err;
ssize_t copied; ssize_t copied;
long timeo = sock_sndtimeo(sk, flags&MSG_DONTWAIT); long timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT);
/* Wait for a connection to finish. */ /* Wait for a connection to finish. */
if ((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) if ((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT))
if((err = wait_for_tcp_connect(sk, 0, &timeo)) != 0) if ((err = wait_for_tcp_connect(sk, 0, &timeo)) != 0)
goto out_err; goto out_err;
clear_bit(SOCK_ASYNC_NOSPACE, &sk->socket->flags); clear_bit(SOCK_ASYNC_NOSPACE, &sk->socket->flags);
...@@ -851,20 +855,19 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size ...@@ -851,20 +855,19 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size
while (psize > 0) { while (psize > 0) {
struct sk_buff *skb = sk->write_queue.prev; struct sk_buff *skb = sk->write_queue.prev;
int offset, size, copy, i; struct page *page = pages[poffset / PAGE_SIZE];
struct page *page; int copy, i;
int offset = poffset % PAGE_SIZE;
int size = min_t(size_t, psize, PAGE_SIZE - offset);
page = pages[poffset/PAGE_SIZE]; if (!tp->send_head || (copy = mss_now - skb->len) <= 0) {
offset = poffset % PAGE_SIZE;
size = min_t(size_t, psize, PAGE_SIZE-offset);
if (tp->send_head==NULL || (copy = mss_now - skb->len) <= 0) {
new_segment: new_segment:
if (!tcp_memory_free(sk)) if (!tcp_memory_free(sk))
goto wait_for_sndbuf; goto wait_for_sndbuf;
skb = tcp_alloc_pskb(sk, 0, tp->mss_cache, sk->allocation); skb = tcp_alloc_pskb(sk, 0, tp->mss_cache,
if (skb == NULL) sk->allocation);
if (!skb)
goto wait_for_memory; goto wait_for_memory;
skb_entail(sk, tp, skb); skb_entail(sk, tp, skb);
...@@ -876,7 +879,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size ...@@ -876,7 +879,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size
i = skb_shinfo(skb)->nr_frags; i = skb_shinfo(skb)->nr_frags;
if (can_coalesce(skb, i, page, offset)) { if (can_coalesce(skb, i, page, offset)) {
skb_shinfo(skb)->frags[i-1].size += copy; skb_shinfo(skb)->frags[i - 1].size += copy;
} else if (i < MAX_SKB_FRAGS) { } else if (i < MAX_SKB_FRAGS) {
get_page(page); get_page(page);
fill_page_desc(skb, i, page, offset, copy); fill_page_desc(skb, i, page, offset, copy);
...@@ -899,7 +902,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size ...@@ -899,7 +902,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size
if (!(psize -= copy)) if (!(psize -= copy))
goto out; goto out;
if (skb->len != mss_now || (flags&MSG_OOB)) if (skb->len != mss_now || (flags & MSG_OOB))
continue; continue;
if (forced_push(tp)) { if (forced_push(tp)) {
...@@ -913,7 +916,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size ...@@ -913,7 +916,7 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size
set_bit(SOCK_NOSPACE, &sk->socket->flags); set_bit(SOCK_NOSPACE, &sk->socket->flags);
wait_for_memory: wait_for_memory:
if (copied) if (copied)
tcp_push(sk, tp, flags&~MSG_MORE, mss_now, 1); tcp_push(sk, tp, flags & ~MSG_MORE, mss_now, 1);
if ((err = wait_for_tcp_memory(sk, &timeo)) != 0) if ((err = wait_for_tcp_memory(sk, &timeo)) != 0)
goto do_error; goto do_error;
...@@ -933,12 +936,13 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size ...@@ -933,12 +936,13 @@ ssize_t do_tcp_sendpages(struct sock *sk, struct page **pages, int poffset, size
return tcp_error(sk, flags, err); return tcp_error(sk, flags, err);
} }
ssize_t tcp_sendpage(struct socket *sock, struct page *page, int offset, size_t size, int flags) ssize_t tcp_sendpage(struct socket *sock, struct page *page, int offset,
size_t size, int flags)
{ {
ssize_t res; ssize_t res;
struct sock *sk = sock->sk; struct sock *sk = sock->sk;
#define TCP_ZC_CSUM_FLAGS (NETIF_F_IP_CSUM|NETIF_F_NO_CSUM|NETIF_F_HW_CSUM) #define TCP_ZC_CSUM_FLAGS (NETIF_F_IP_CSUM | NETIF_F_NO_CSUM | NETIF_F_HW_CSUM)
if (!(sk->route_caps & NETIF_F_SG) || if (!(sk->route_caps & NETIF_F_SG) ||
!(sk->route_caps & TCP_ZC_CSUM_FLAGS)) !(sk->route_caps & TCP_ZC_CSUM_FLAGS))
...@@ -957,14 +961,14 @@ ssize_t tcp_sendpage(struct socket *sock, struct page *page, int offset, size_t ...@@ -957,14 +961,14 @@ ssize_t tcp_sendpage(struct socket *sock, struct page *page, int offset, size_t
#define TCP_PAGE(sk) (tcp_sk(sk)->sndmsg_page) #define TCP_PAGE(sk) (tcp_sk(sk)->sndmsg_page)
#define TCP_OFF(sk) (tcp_sk(sk)->sndmsg_off) #define TCP_OFF(sk) (tcp_sk(sk)->sndmsg_off)
static inline int static inline int tcp_copy_to_page(struct sock *sk, char *from,
tcp_copy_to_page(struct sock *sk, char *from, struct sk_buff *skb, struct sk_buff *skb, struct page *page,
struct page *page, int off, int copy) int off, int copy)
{ {
int err = 0; int err = 0;
unsigned int csum; unsigned int csum;
csum = csum_and_copy_from_user(from, page_address(page)+off, csum = csum_and_copy_from_user(from, page_address(page) + off,
copy, 0, &err); copy, 0, &err);
if (!err) { if (!err) {
if (skb->ip_summed == CHECKSUM_NONE) if (skb->ip_summed == CHECKSUM_NONE)
...@@ -978,8 +982,7 @@ tcp_copy_to_page(struct sock *sk, char *from, struct sk_buff *skb, ...@@ -978,8 +982,7 @@ tcp_copy_to_page(struct sock *sk, char *from, struct sk_buff *skb,
return err; return err;
} }
static inline int static inline int skb_add_data(struct sk_buff *skb, char *from, int copy)
skb_add_data(struct sk_buff *skb, char *from, int copy)
{ {
int err = 0; int err = 0;
unsigned int csum; unsigned int csum;
...@@ -1000,10 +1003,11 @@ static inline int select_size(struct sock *sk, struct tcp_opt *tp) ...@@ -1000,10 +1003,11 @@ static inline int select_size(struct sock *sk, struct tcp_opt *tp)
{ {
int tmp = tp->mss_cache; int tmp = tp->mss_cache;
if (sk->route_caps&NETIF_F_SG) { if (sk->route_caps & NETIF_F_SG) {
int pgbreak = SKB_MAX_HEAD(MAX_TCP_HEADER); int pgbreak = SKB_MAX_HEAD(MAX_TCP_HEADER);
if (tmp >= pgbreak && tmp <= pgbreak + (MAX_SKB_FRAGS-1)*PAGE_SIZE) if (tmp >= pgbreak &&
tmp <= pgbreak + (MAX_SKB_FRAGS - 1) * PAGE_SIZE)
tmp = pgbreak; tmp = pgbreak;
} }
return tmp; return tmp;
...@@ -1023,11 +1027,11 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1023,11 +1027,11 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
TCP_CHECK_TIMER(sk); TCP_CHECK_TIMER(sk);
flags = msg->msg_flags; flags = msg->msg_flags;
timeo = sock_sndtimeo(sk, flags&MSG_DONTWAIT); timeo = sock_sndtimeo(sk, flags & MSG_DONTWAIT);
/* Wait for a connection to finish. */ /* Wait for a connection to finish. */
if ((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) if ((1 << sk->state) & ~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT))
if((err = wait_for_tcp_connect(sk, flags, &timeo)) != 0) if ((err = wait_for_tcp_connect(sk, flags, &timeo)) != 0)
goto out_err; goto out_err;
/* This should be in poll */ /* This should be in poll */
...@@ -1041,12 +1045,12 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1041,12 +1045,12 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
copied = 0; copied = 0;
err = -EPIPE; err = -EPIPE;
if (sk->err || (sk->shutdown&SEND_SHUTDOWN)) if (sk->err || (sk->shutdown & SEND_SHUTDOWN))
goto do_error; goto do_error;
while (--iovlen >= 0) { while (--iovlen >= 0) {
int seglen=iov->iov_len; int seglen = iov->iov_len;
unsigned char * from=iov->iov_base; unsigned char *from = iov->iov_base;
iov++; iov++;
...@@ -1055,7 +1059,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1055,7 +1059,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
skb = sk->write_queue.prev; skb = sk->write_queue.prev;
if (tp->send_head == NULL || if (!tp->send_head ||
(copy = mss_now - skb->len) <= 0) { (copy = mss_now - skb->len) <= 0) {
new_segment: new_segment:
...@@ -1065,8 +1069,9 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1065,8 +1069,9 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
if (!tcp_memory_free(sk)) if (!tcp_memory_free(sk))
goto wait_for_sndbuf; goto wait_for_sndbuf;
skb = tcp_alloc_pskb(sk, select_size(sk, tp), 0, sk->allocation); skb = tcp_alloc_pskb(sk, select_size(sk, tp),
if (skb == NULL) 0, sk->allocation);
if (!skb)
goto wait_for_memory; goto wait_for_memory;
skb_entail(sk, tp, skb); skb_entail(sk, tp, skb);
...@@ -1090,22 +1095,26 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1090,22 +1095,26 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
struct page *page = TCP_PAGE(sk); struct page *page = TCP_PAGE(sk);
int off = TCP_OFF(sk); int off = TCP_OFF(sk);
if (can_coalesce(skb, i, page, off) && off != PAGE_SIZE) { if (can_coalesce(skb, i, page, off) &&
/* We can extend the last page fragment. */ off != PAGE_SIZE) {
/* We can extend the last page
* fragment. */
merge = 1; merge = 1;
} else if (i == MAX_SKB_FRAGS || } else if (i == MAX_SKB_FRAGS ||
(i == 0 && !(sk->route_caps&NETIF_F_SG))) { (!i &&
!(sk->route_caps & NETIF_F_SG))) {
/* Need to add new fragment and cannot /* Need to add new fragment and cannot
* do this because interface is non-SG, * do this because interface is non-SG,
* or because all the page slots are busy. * or because all the page slots are
*/ * busy. */
tcp_mark_push(tp, skb); tcp_mark_push(tp, skb);
goto new_segment; goto new_segment;
} else if (page) { } else if (page) {
/* If page is cached, align /* If page is cached, align
* offset to L1 cache boundary * offset to L1 cache boundary
*/ */
off = (off+L1_CACHE_BYTES-1)&~(L1_CACHE_BYTES-1); off = (off + L1_CACHE_BYTES - 1) &
~(L1_CACHE_BYTES - 1);
if (off == PAGE_SIZE) { if (off == PAGE_SIZE) {
put_page(page); put_page(page);
TCP_PAGE(sk) = page = NULL; TCP_PAGE(sk) = page = NULL;
...@@ -1114,21 +1123,23 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1114,21 +1123,23 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
if (!page) { if (!page) {
/* Allocate new cache page. */ /* Allocate new cache page. */
if (!(page=tcp_alloc_page(sk))) if (!(page = tcp_alloc_page(sk)))
goto wait_for_memory; goto wait_for_memory;
off = 0; off = 0;
} }
if (copy > PAGE_SIZE-off) if (copy > PAGE_SIZE - off)
copy = PAGE_SIZE-off; copy = PAGE_SIZE - off;
/* Time to copy data. We are close to the end! */ /* Time to copy data. We are close to
err = tcp_copy_to_page(sk, from, skb, page, off, copy); * the end! */
err = tcp_copy_to_page(sk, from, skb, page,
off, copy);
if (err) { if (err) {
/* If this page was new, give it to the /* If this page was new, give it to the
* socket so it does not get leaked. * socket so it does not get leaked.
*/ */
if (TCP_PAGE(sk) == NULL) { if (!TCP_PAGE(sk)) {
TCP_PAGE(sk) = page; TCP_PAGE(sk) = page;
TCP_OFF(sk) = 0; TCP_OFF(sk) = 0;
} }
...@@ -1137,7 +1148,8 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1137,7 +1148,8 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
/* Update the skb. */ /* Update the skb. */
if (merge) { if (merge) {
skb_shinfo(skb)->frags[i-1].size += copy; skb_shinfo(skb)->frags[i - 1].size +=
copy;
} else { } else {
fill_page_desc(skb, i, page, off, copy); fill_page_desc(skb, i, page, off, copy);
if (TCP_PAGE(sk)) { if (TCP_PAGE(sk)) {
...@@ -1148,7 +1160,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1148,7 +1160,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
} }
} }
TCP_OFF(sk) = off+copy; TCP_OFF(sk) = off + copy;
} }
if (!copied) if (!copied)
...@@ -1161,7 +1173,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1161,7 +1173,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
copied += copy; copied += copy;
seglen -= copy; seglen -= copy;
if (skb->len != mss_now || (flags&MSG_OOB)) if (skb->len != mss_now || (flags & MSG_OOB))
continue; continue;
if (forced_push(tp)) { if (forced_push(tp)) {
...@@ -1175,7 +1187,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1175,7 +1187,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
set_bit(SOCK_NOSPACE, &sk->socket->flags); set_bit(SOCK_NOSPACE, &sk->socket->flags);
wait_for_memory: wait_for_memory:
if (copied) if (copied)
tcp_push(sk, tp, flags&~MSG_MORE, mss_now, 1); tcp_push(sk, tp, flags & ~MSG_MORE, mss_now, 1);
if ((err = wait_for_tcp_memory(sk, &timeo)) != 0) if ((err = wait_for_tcp_memory(sk, &timeo)) != 0)
goto do_error; goto do_error;
...@@ -1192,7 +1204,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1192,7 +1204,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
return copied; return copied;
do_fault: do_fault:
if (skb->len == 0) { if (!skb->len) {
if (tp->send_head == skb) if (tp->send_head == skb)
tp->send_head = NULL; tp->send_head = NULL;
__skb_unlink(skb, skb->list); __skb_unlink(skb, skb->list);
...@@ -1214,7 +1226,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) ...@@ -1214,7 +1226,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size)
* this, no blocking and very strange errors 8) * this, no blocking and very strange errors 8)
*/ */
static int tcp_recv_urg(struct sock * sk, long timeo, static int tcp_recv_urg(struct sock *sk, long timeo,
struct msghdr *msg, int len, int flags, struct msghdr *msg, int len, int flags,
int *addr_len) int *addr_len)
{ {
...@@ -1224,7 +1236,7 @@ static int tcp_recv_urg(struct sock * sk, long timeo, ...@@ -1224,7 +1236,7 @@ static int tcp_recv_urg(struct sock * sk, long timeo,
if (sk->urginline || !tp->urg_data || tp->urg_data == TCP_URG_READ) if (sk->urginline || !tp->urg_data || tp->urg_data == TCP_URG_READ)
return -EINVAL; /* Yes this is right ! */ return -EINVAL; /* Yes this is right ! */
if (sk->state==TCP_CLOSE && !sk->done) if (sk->state == TCP_CLOSE && !sk->done)
return -ENOTCONN; return -ENOTCONN;
if (tp->urg_data & TCP_URG_VALID) { if (tp->urg_data & TCP_URG_VALID) {
...@@ -1235,14 +1247,14 @@ static int tcp_recv_urg(struct sock * sk, long timeo, ...@@ -1235,14 +1247,14 @@ static int tcp_recv_urg(struct sock * sk, long timeo,
tp->urg_data = TCP_URG_READ; tp->urg_data = TCP_URG_READ;
/* Read urgent data. */ /* Read urgent data. */
msg->msg_flags|=MSG_OOB; msg->msg_flags |= MSG_OOB;
if(len>0) { if (len > 0) {
if (!(flags & MSG_TRUNC)) if (!(flags & MSG_TRUNC))
err = memcpy_toiovec(msg->msg_iov, &c, 1); err = memcpy_toiovec(msg->msg_iov, &c, 1);
len = 1; len = 1;
} else } else
msg->msg_flags|=MSG_TRUNC; msg->msg_flags |= MSG_TRUNC;
return err ? -EFAULT : len; return err ? -EFAULT : len;
} }
...@@ -1265,7 +1277,7 @@ static int tcp_recv_urg(struct sock * sk, long timeo, ...@@ -1265,7 +1277,7 @@ static int tcp_recv_urg(struct sock * sk, long timeo,
* socket locked so that the sk_buff queue operation is ok. * socket locked so that the sk_buff queue operation is ok.
*/ */
static inline void tcp_eat_skb(struct sock *sk, struct sk_buff * skb) static inline void tcp_eat_skb(struct sock *sk, struct sk_buff *skb)
{ {
__skb_unlink(skb, &sk->receive_queue); __skb_unlink(skb, &sk->receive_queue);
__kfree_skb(skb); __kfree_skb(skb);
...@@ -1285,27 +1297,25 @@ static void cleanup_rbuf(struct sock *sk, int copied) ...@@ -1285,27 +1297,25 @@ static void cleanup_rbuf(struct sock *sk, int copied)
#if TCP_DEBUG #if TCP_DEBUG
struct sk_buff *skb = skb_peek(&sk->receive_queue); struct sk_buff *skb = skb_peek(&sk->receive_queue);
BUG_TRAP(skb==NULL || before(tp->copied_seq, TCP_SKB_CB(skb)->end_seq)); BUG_TRAP(!skb || before(tp->copied_seq, TCP_SKB_CB(skb)->end_seq));
#endif #endif
if (tcp_ack_scheduled(tp)) { if (tcp_ack_scheduled(tp)) {
/* Delayed ACKs frequently hit locked sockets during bulk receive. */ /* Delayed ACKs frequently hit locked sockets during bulk
if (tp->ack.blocked * receive. */
if (tp->ack.blocked ||
/* Once-per-two-segments ACK was not sent by tcp_input.c */ /* Once-per-two-segments ACK was not sent by tcp_input.c */
|| tp->rcv_nxt - tp->rcv_wup > tp->ack.rcv_mss tp->rcv_nxt - tp->rcv_wup > tp->ack.rcv_mss ||
/* /*
* If this read emptied read buffer, we send ACK, if * If this read emptied read buffer, we send ACK, if
* connection is not bidirectional, user drained * connection is not bidirectional, user drained
* receive buffer and there was a small segment * receive buffer and there was a small segment
* in queue. * in queue.
*/ */
|| (copied > 0 && (copied > 0 && (tp->ack.pending & TCP_ACK_PUSHED) &&
(tp->ack.pending&TCP_ACK_PUSHED) && !tp->ack.pingpong && !atomic_read(&sk->rmem_alloc)))
!tp->ack.pingpong &&
atomic_read(&sk->rmem_alloc) == 0)) {
time_to_ack = 1; time_to_ack = 1;
} }
}
/* We send an ACK if we can now advertise a non-zero window /* We send an ACK if we can now advertise a non-zero window
* which has been raised "significantly". * which has been raised "significantly".
...@@ -1313,7 +1323,7 @@ static void cleanup_rbuf(struct sock *sk, int copied) ...@@ -1313,7 +1323,7 @@ static void cleanup_rbuf(struct sock *sk, int copied)
* Even if window raised up to infinity, do not send window open ACK * Even if window raised up to infinity, do not send window open ACK
* in states, where we will not receive more. It is useless. * in states, where we will not receive more. It is useless.
*/ */
if(copied > 0 && !time_to_ack && !(sk->shutdown&RCV_SHUTDOWN)) { if (copied > 0 && !time_to_ack && !(sk->shutdown & RCV_SHUTDOWN)) {
__u32 rcv_window_now = tcp_receive_window(tp); __u32 rcv_window_now = tcp_receive_window(tp);
/* Optimize, __tcp_select_window() is not cheap. */ /* Optimize, __tcp_select_window() is not cheap. */
...@@ -1325,7 +1335,7 @@ static void cleanup_rbuf(struct sock *sk, int copied) ...@@ -1325,7 +1335,7 @@ static void cleanup_rbuf(struct sock *sk, int copied)
* We can advertise it now, if it is not less than current one. * We can advertise it now, if it is not less than current one.
* "Lots" means "at least twice" here. * "Lots" means "at least twice" here.
*/ */
if(new_window && new_window >= 2*rcv_window_now) if (new_window && new_window >= 2 * rcv_window_now)
time_to_ack = 1; time_to_ack = 1;
} }
} }
...@@ -1366,9 +1376,11 @@ static void tcp_prequeue_process(struct sock *sk) ...@@ -1366,9 +1376,11 @@ static void tcp_prequeue_process(struct sock *sk)
struct sk_buff *skb; struct sk_buff *skb;
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
net_statistics[smp_processor_id()*2+1].TCPPrequeued += skb_queue_len(&tp->ucopy.prequeue); net_statistics[smp_processor_id() * 2 + 1].TCPPrequeued +=
skb_queue_len(&tp->ucopy.prequeue);
/* RX process wants to run with disabled BHs, though it is not necessary */ /* RX process wants to run with disabled BHs, though it is not
* necessary */
local_bh_disable(); local_bh_disable();
while ((skb = __skb_dequeue(&tp->ucopy.prequeue)) != NULL) while ((skb = __skb_dequeue(&tp->ucopy.prequeue)) != NULL)
sk->backlog_rcv(sk, skb); sk->backlog_rcv(sk, skb);
...@@ -1378,8 +1390,7 @@ static void tcp_prequeue_process(struct sock *sk) ...@@ -1378,8 +1390,7 @@ static void tcp_prequeue_process(struct sock *sk)
tp->ucopy.memory = 0; tp->ucopy.memory = 0;
} }
static inline static inline struct sk_buff *tcp_recv_skb(struct sock *sk, u32 seq, u32 *off)
struct sk_buff *tcp_recv_skb(struct sock *sk, u32 seq, u32 *off)
{ {
struct sk_buff *skb; struct sk_buff *skb;
u32 offset; u32 offset;
...@@ -1500,7 +1511,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1500,7 +1511,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
target = sock_rcvlowat(sk, flags & MSG_WAITALL, len); target = sock_rcvlowat(sk, flags & MSG_WAITALL, len);
do { do {
struct sk_buff * skb; struct sk_buff *skb;
u32 offset; u32 offset;
/* Are we at urgent data? Stop if we have read anything. */ /* Are we at urgent data? Stop if we have read anything. */
...@@ -1529,8 +1540,8 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1529,8 +1540,8 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
* shouldn't happen. * shouldn't happen.
*/ */
if (before(*seq, TCP_SKB_CB(skb)->seq)) { if (before(*seq, TCP_SKB_CB(skb)->seq)) {
printk(KERN_INFO "recvmsg bug: copied %X seq %X\n", printk(KERN_INFO "recvmsg bug: copied %X "
*seq, TCP_SKB_CB(skb)->seq); "seq %X\n", *seq, TCP_SKB_CB(skb)->seq);
break; break;
} }
offset = *seq - TCP_SKB_CB(skb)->seq; offset = *seq - TCP_SKB_CB(skb)->seq;
...@@ -1540,13 +1551,13 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1540,13 +1551,13 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
goto found_ok_skb; goto found_ok_skb;
if (skb->h.th->fin) if (skb->h.th->fin)
goto found_fin_ok; goto found_fin_ok;
BUG_TRAP(flags&MSG_PEEK); BUG_TRAP(flags & MSG_PEEK);
skb = skb->next; skb = skb->next;
} while (skb != (struct sk_buff *)&sk->receive_queue); } while (skb != (struct sk_buff *)&sk->receive_queue);
/* Well, if we have backlog, try to process it now yet. */ /* Well, if we have backlog, try to process it now yet. */
if (copied >= target && sk->backlog.tail == NULL) if (copied >= target && !sk->backlog.tail)
break; break;
if (copied) { if (copied) {
...@@ -1589,7 +1600,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1589,7 +1600,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
if (tp->ucopy.task == user_recv) { if (tp->ucopy.task == user_recv) {
/* Install new reader */ /* Install new reader */
if (user_recv == NULL && !(flags&(MSG_TRUNC|MSG_PEEK))) { if (!user_recv && !(flags & (MSG_TRUNC | MSG_PEEK))) {
user_recv = current; user_recv = current;
tp->ucopy.task = user_recv; tp->ucopy.task = user_recv;
tp->ucopy.iov = msg->msg_iov; tp->ucopy.iov = msg->msg_iov;
...@@ -1597,7 +1608,8 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1597,7 +1608,8 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
tp->ucopy.len = len; tp->ucopy.len = len;
BUG_TRAP(tp->copied_seq == tp->rcv_nxt || (flags&(MSG_PEEK|MSG_TRUNC))); BUG_TRAP(tp->copied_seq == tp->rcv_nxt ||
(flags & (MSG_PEEK | MSG_TRUNC)));
/* Ugly... If prequeue is not empty, we have to /* Ugly... If prequeue is not empty, we have to
* process it before releasing socket, otherwise * process it before releasing socket, otherwise
...@@ -1613,7 +1625,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1613,7 +1625,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
* *
* Each queue can be processed only if the next ones * Each queue can be processed only if the next ones
* are empty. At this point we have empty receive_queue. * are empty. At this point we have empty receive_queue.
* But prequeue _can_ be not empty after second iteration, * But prequeue _can_ be not empty after 2nd iteration,
* when we jumped to start of loop because backlog * when we jumped to start of loop because backlog
* processing added something to receive_queue. * processing added something to receive_queue.
* We cannot release_sock(), because backlog contains * We cannot release_sock(), because backlog contains
...@@ -1645,7 +1657,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1645,7 +1657,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
/* __ Restore normal policy in scheduler __ */ /* __ Restore normal policy in scheduler __ */
if ((chunk = len - tp->ucopy.len) != 0) { if ((chunk = len - tp->ucopy.len) != 0) {
net_statistics[smp_processor_id()*2+1].TCPDirectCopyFromBacklog += chunk; net_statistics[smp_processor_id() * 2 + 1].TCPDirectCopyFromBacklog += chunk;
len -= chunk; len -= chunk;
copied += chunk; copied += chunk;
} }
...@@ -1656,7 +1668,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1656,7 +1668,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
tcp_prequeue_process(sk); tcp_prequeue_process(sk);
if ((chunk = len - tp->ucopy.len) != 0) { if ((chunk = len - tp->ucopy.len) != 0) {
net_statistics[smp_processor_id()*2+1].TCPDirectCopyFromPrequeue += chunk; net_statistics[smp_processor_id() * 2 + 1].TCPDirectCopyFromPrequeue += chunk;
len -= chunk; len -= chunk;
copied += chunk; copied += chunk;
} }
...@@ -1687,8 +1699,9 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1687,8 +1699,9 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
} }
} }
if (!(flags&MSG_TRUNC)) { if (!(flags & MSG_TRUNC)) {
err = skb_copy_datagram_iovec(skb, offset, msg->msg_iov, used); err = skb_copy_datagram_iovec(skb, offset,
msg->msg_iov, used);
if (err) { if (err) {
/* Exception. Bailout! */ /* Exception. Bailout! */
if (!copied) if (!copied)
...@@ -1702,7 +1715,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1702,7 +1715,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
len -= used; len -= used;
skip_copy: skip_copy:
if (tp->urg_data && after(tp->copied_seq,tp->urg_seq)) { if (tp->urg_data && after(tp->copied_seq, tp->urg_seq)) {
tp->urg_data = 0; tp->urg_data = 0;
tcp_fast_path_check(sk, tp); tcp_fast_path_check(sk, tp);
} }
...@@ -1732,7 +1745,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg, ...@@ -1732,7 +1745,7 @@ int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
tcp_prequeue_process(sk); tcp_prequeue_process(sk);
if (copied > 0 && (chunk = len - tp->ucopy.len) != 0) { if (copied > 0 && (chunk = len - tp->ucopy.len) != 0) {
net_statistics[smp_processor_id()*2+1].TCPDirectCopyFromPrequeue += chunk; net_statistics[smp_processor_id() * 2 + 1].TCPDirectCopyFromPrequeue += chunk;
len -= chunk; len -= chunk;
copied += chunk; copied += chunk;
} }
...@@ -1788,12 +1801,12 @@ static unsigned char new_state[16] = { ...@@ -1788,12 +1801,12 @@ static unsigned char new_state[16] = {
static int tcp_close_state(struct sock *sk) static int tcp_close_state(struct sock *sk)
{ {
int next = (int) new_state[sk->state]; int next = (int)new_state[sk->state];
int ns = (next & TCP_STATE_MASK); int ns = next & TCP_STATE_MASK;
tcp_set_state(sk, ns); tcp_set_state(sk, ns);
return (next & TCP_ACTION_FIN); return next & TCP_ACTION_FIN;
} }
/* /*
...@@ -1812,7 +1825,8 @@ void tcp_shutdown(struct sock *sk, int how) ...@@ -1812,7 +1825,8 @@ void tcp_shutdown(struct sock *sk, int how)
/* If we've already sent a FIN, or it's a closed state, skip this. */ /* If we've already sent a FIN, or it's a closed state, skip this. */
if ((1 << sk->state) & if ((1 << sk->state) &
(TCPF_ESTABLISHED|TCPF_SYN_SENT|TCPF_SYN_RECV|TCPF_CLOSE_WAIT)) { (TCPF_ESTABLISHED | TCPF_SYN_SENT |
TCPF_SYN_RECV | TCPF_CLOSE_WAIT)) {
/* Clear out any half completed packets. FIN if needed. */ /* Clear out any half completed packets. FIN if needed. */
if (tcp_close_state(sk)) if (tcp_close_state(sk))
tcp_send_fin(sk); tcp_send_fin(sk);
...@@ -1824,9 +1838,10 @@ void tcp_shutdown(struct sock *sk, int how) ...@@ -1824,9 +1838,10 @@ void tcp_shutdown(struct sock *sk, int how)
* Return 1 if we still have things to send in our buffers. * Return 1 if we still have things to send in our buffers.
*/ */
static inline int closing(struct sock * sk) static inline int closing(struct sock *sk)
{ {
return ((1 << sk->state) & (TCPF_FIN_WAIT1|TCPF_CLOSING|TCPF_LAST_ACK)); return (1 << sk->state) &
(TCPF_FIN_WAIT1 | TCPF_CLOSING | TCPF_LAST_ACK);
} }
static __inline__ void tcp_kill_sk_queues(struct sock *sk) static __inline__ void tcp_kill_sk_queues(struct sock *sk)
...@@ -1843,8 +1858,8 @@ static __inline__ void tcp_kill_sk_queues(struct sock *sk) ...@@ -1843,8 +1858,8 @@ static __inline__ void tcp_kill_sk_queues(struct sock *sk)
/* Account for returned memory. */ /* Account for returned memory. */
tcp_mem_reclaim(sk); tcp_mem_reclaim(sk);
BUG_TRAP(sk->wmem_queued == 0); BUG_TRAP(!sk->wmem_queued);
BUG_TRAP(sk->forward_alloc == 0); BUG_TRAP(!sk->forward_alloc);
/* It is _impossible_ for the backlog to contain anything /* It is _impossible_ for the backlog to contain anything
* when we get here. All user references to this socket * when we get here. All user references to this socket
...@@ -1860,11 +1875,11 @@ static __inline__ void tcp_kill_sk_queues(struct sock *sk) ...@@ -1860,11 +1875,11 @@ static __inline__ void tcp_kill_sk_queues(struct sock *sk)
*/ */
void tcp_destroy_sock(struct sock *sk) void tcp_destroy_sock(struct sock *sk)
{ {
BUG_TRAP(sk->state==TCP_CLOSE); BUG_TRAP(sk->state == TCP_CLOSE);
BUG_TRAP(sk->dead); BUG_TRAP(sk->dead);
/* It cannot be in hash table! */ /* It cannot be in hash table! */
BUG_TRAP(sk->pprev==NULL); BUG_TRAP(!sk->pprev);
/* If it has not 0 inet_sk(sk)->num, it must be bound */ /* If it has not 0 inet_sk(sk)->num, it must be bound */
BUG_TRAP(!inet_sk(sk)->num || sk->prev); BUG_TRAP(!inet_sk(sk)->num || sk->prev);
...@@ -1883,7 +1898,8 @@ void tcp_destroy_sock(struct sock *sk) ...@@ -1883,7 +1898,8 @@ void tcp_destroy_sock(struct sock *sk)
#ifdef INET_REFCNT_DEBUG #ifdef INET_REFCNT_DEBUG
if (atomic_read(&sk->refcnt) != 1) { if (atomic_read(&sk->refcnt) != 1) {
printk(KERN_DEBUG "Destruction TCP %p delayed, c=%d\n", sk, atomic_read(&sk->refcnt)); printk(KERN_DEBUG "Destruction TCP %p delayed, c=%d\n",
sk, atomic_read(&sk->refcnt));
} }
#endif #endif
...@@ -1899,7 +1915,7 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -1899,7 +1915,7 @@ void tcp_close(struct sock *sk, long timeout)
lock_sock(sk); lock_sock(sk);
sk->shutdown = SHUTDOWN_MASK; sk->shutdown = SHUTDOWN_MASK;
if(sk->state == TCP_LISTEN) { if (sk->state == TCP_LISTEN) {
tcp_set_state(sk, TCP_CLOSE); tcp_set_state(sk, TCP_CLOSE);
/* Special case. */ /* Special case. */
...@@ -1912,8 +1928,9 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -1912,8 +1928,9 @@ void tcp_close(struct sock *sk, long timeout)
* descriptor close, not protocol-sourced closes, because the * descriptor close, not protocol-sourced closes, because the
* reader process may not have drained the data yet! * reader process may not have drained the data yet!
*/ */
while((skb=__skb_dequeue(&sk->receive_queue))!=NULL) { while ((skb = __skb_dequeue(&sk->receive_queue)) != NULL) {
u32 len = TCP_SKB_CB(skb)->end_seq - TCP_SKB_CB(skb)->seq - skb->h.th->fin; u32 len = TCP_SKB_CB(skb)->end_seq - TCP_SKB_CB(skb)->seq -
skb->h.th->fin;
data_was_unread += len; data_was_unread += len;
__kfree_skb(skb); __kfree_skb(skb);
} }
...@@ -1929,12 +1946,12 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -1929,12 +1946,12 @@ void tcp_close(struct sock *sk, long timeout)
* the FTP client, wheee... Note: timeout is always zero * the FTP client, wheee... Note: timeout is always zero
* in such a case. * in such a case.
*/ */
if(data_was_unread != 0) { if (data_was_unread) {
/* Unread data was tossed, zap the connection. */ /* Unread data was tossed, zap the connection. */
NET_INC_STATS_USER(TCPAbortOnClose); NET_INC_STATS_USER(TCPAbortOnClose);
tcp_set_state(sk, TCP_CLOSE); tcp_set_state(sk, TCP_CLOSE);
tcp_send_active_reset(sk, GFP_KERNEL); tcp_send_active_reset(sk, GFP_KERNEL);
} else if (sk->linger && sk->lingertime==0) { } else if (sk->linger && !sk->lingertime) {
/* Check zero linger _after_ checking for unread data. */ /* Check zero linger _after_ checking for unread data. */
sk->prot->disconnect(sk, 0); sk->prot->disconnect(sk, 0);
NET_INC_STATS_USER(TCPAbortOnData); NET_INC_STATS_USER(TCPAbortOnData);
...@@ -1996,7 +2013,7 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -1996,7 +2013,7 @@ void tcp_close(struct sock *sk, long timeout)
*/ */
local_bh_disable(); local_bh_disable();
bh_lock_sock(sk); bh_lock_sock(sk);
BUG_TRAP(sk->lock.users==0); BUG_TRAP(!sk->lock.users);
sock_hold(sk); sock_hold(sk);
sock_orphan(sk); sock_orphan(sk);
...@@ -2039,7 +2056,8 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -2039,7 +2056,8 @@ void tcp_close(struct sock *sk, long timeout)
(sk->wmem_queued > SOCK_MIN_SNDBUF && (sk->wmem_queued > SOCK_MIN_SNDBUF &&
atomic_read(&tcp_memory_allocated) > sysctl_tcp_mem[2])) { atomic_read(&tcp_memory_allocated) > sysctl_tcp_mem[2])) {
if (net_ratelimit()) if (net_ratelimit())
printk(KERN_INFO "TCP: too many of orphaned sockets\n"); printk(KERN_INFO "TCP: too many of orphaned "
"sockets\n");
tcp_set_state(sk, TCP_CLOSE); tcp_set_state(sk, TCP_CLOSE);
tcp_send_active_reset(sk, GFP_ATOMIC); tcp_send_active_reset(sk, GFP_ATOMIC);
NET_INC_STATS_BH(TCPAbortOnMemory); NET_INC_STATS_BH(TCPAbortOnMemory);
...@@ -2061,9 +2079,9 @@ void tcp_close(struct sock *sk, long timeout) ...@@ -2061,9 +2079,9 @@ void tcp_close(struct sock *sk, long timeout)
extern __inline__ int tcp_need_reset(int state) extern __inline__ int tcp_need_reset(int state)
{ {
return ((1 << state) & return (1 << state) &
(TCPF_ESTABLISHED|TCPF_CLOSE_WAIT|TCPF_FIN_WAIT1| (TCPF_ESTABLISHED | TCPF_CLOSE_WAIT | TCPF_FIN_WAIT1 |
TCPF_FIN_WAIT2|TCPF_SYN_RECV)); TCPF_FIN_WAIT2 | TCPF_SYN_RECV);
} }
int tcp_disconnect(struct sock *sk, int flags) int tcp_disconnect(struct sock *sk, int flags)
...@@ -2082,7 +2100,7 @@ int tcp_disconnect(struct sock *sk, int flags) ...@@ -2082,7 +2100,7 @@ int tcp_disconnect(struct sock *sk, int flags)
tcp_listen_stop(sk); tcp_listen_stop(sk);
} else if (tcp_need_reset(old_state) || } else if (tcp_need_reset(old_state) ||
(tp->snd_nxt != tp->write_seq && (tp->snd_nxt != tp->write_seq &&
(1<<old_state)&(TCPF_CLOSING|TCPF_LAST_ACK))) { (1 << old_state) & (TCPF_CLOSING | TCPF_LAST_ACK))) {
/* The last check adjusts for discrepance of Linux wrt. RFC /* The last check adjusts for discrepance of Linux wrt. RFC
* states * states
*/ */
...@@ -2098,7 +2116,7 @@ int tcp_disconnect(struct sock *sk, int flags) ...@@ -2098,7 +2116,7 @@ int tcp_disconnect(struct sock *sk, int flags)
inet->dport = 0; inet->dport = 0;
if (!(sk->userlocks&SOCK_BINDADDR_LOCK)) { if (!(sk->userlocks & SOCK_BINDADDR_LOCK)) {
inet->rcv_saddr = inet->saddr = 0; inet->rcv_saddr = inet->saddr = 0;
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
if (sk->family == PF_INET6) { if (sk->family == PF_INET6) {
...@@ -2113,7 +2131,7 @@ int tcp_disconnect(struct sock *sk, int flags) ...@@ -2113,7 +2131,7 @@ int tcp_disconnect(struct sock *sk, int flags)
sk->shutdown = 0; sk->shutdown = 0;
sk->done = 0; sk->done = 0;
tp->srtt = 0; tp->srtt = 0;
if ((tp->write_seq += tp->max_window+2) == 0) if ((tp->write_seq += tp->max_window + 2) == 0)
tp->write_seq = 1; tp->write_seq = 1;
tp->backoff = 0; tp->backoff = 0;
tp->snd_cwnd = 2; tp->snd_cwnd = 2;
...@@ -2139,7 +2157,7 @@ int tcp_disconnect(struct sock *sk, int flags) ...@@ -2139,7 +2157,7 @@ int tcp_disconnect(struct sock *sk, int flags)
* Wait for an incoming connection, avoid race * Wait for an incoming connection, avoid race
* conditions. This must be called with the socket locked. * conditions. This must be called with the socket locked.
*/ */
static int wait_for_connect(struct sock * sk, long timeo) static int wait_for_connect(struct sock *sk, long timeo)
{ {
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
DECLARE_WAITQUEUE(wait, current); DECLARE_WAITQUEUE(wait, current);
...@@ -2163,7 +2181,7 @@ static int wait_for_connect(struct sock * sk, long timeo) ...@@ -2163,7 +2181,7 @@ static int wait_for_connect(struct sock * sk, long timeo)
for (;;) { for (;;) {
current->state = TASK_INTERRUPTIBLE; current->state = TASK_INTERRUPTIBLE;
release_sock(sk); release_sock(sk);
if (tp->accept_queue == NULL) if (!tp->accept_queue)
timeo = schedule_timeout(timeo); timeo = schedule_timeout(timeo);
lock_sock(sk); lock_sock(sk);
err = 0; err = 0;
...@@ -2238,7 +2256,6 @@ struct sock *tcp_accept(struct sock *sk, int flags, int *err) ...@@ -2238,7 +2256,6 @@ struct sock *tcp_accept(struct sock *sk, int flags, int *err)
/* /*
* Socket option code for TCP. * Socket option code for TCP.
*/ */
int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
int optlen) int optlen)
{ {
...@@ -2250,7 +2267,7 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2250,7 +2267,7 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
return tp->af_specific->setsockopt(sk, level, optname, return tp->af_specific->setsockopt(sk, level, optname,
optval, optlen); optval, optlen);
if(optlen<sizeof(int)) if (optlen < sizeof(int))
return -EINVAL; return -EINVAL;
if (get_user(val, (int *)optval)) if (get_user(val, (int *)optval))
...@@ -2258,13 +2275,12 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2258,13 +2275,12 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
lock_sock(sk); lock_sock(sk);
switch(optname) { switch (optname) {
case TCP_MAXSEG: case TCP_MAXSEG:
/* values greater than interface MTU won't take effect. however at /* Values greater than interface MTU won't take effect. However
* the point when this call is done we typically don't yet know * at the point when this call is done we typically don't yet
* which interface is going to be used * know which interface is going to be used */
*/ if (val < 8 || val > MAX_TCP_WINDOW) {
if(val < 8 || val > MAX_TCP_WINDOW) {
err = -EINVAL; err = -EINVAL;
break; break;
} }
...@@ -2279,7 +2295,7 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2279,7 +2295,7 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
err = -EINVAL; err = -EINVAL;
break; break;
} }
tp->nonagle = (val == 0) ? 0 : 1; tp->nonagle = !val ? 0 : 1;
if (val) if (val)
tcp_push_pending_frames(sk, tp); tcp_push_pending_frames(sk, tp);
break; break;
...@@ -2314,7 +2330,8 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2314,7 +2330,8 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
err = -EINVAL; err = -EINVAL;
else { else {
tp->keepalive_time = val * HZ; tp->keepalive_time = val * HZ;
if (sk->keepopen && !((1<<sk->state)&(TCPF_CLOSE|TCPF_LISTEN))) { if (sk->keepopen &&
!((1 << sk->state) & (TCPF_CLOSE | TCPF_LISTEN))) {
__u32 elapsed = tcp_time_stamp - tp->rcv_tstamp; __u32 elapsed = tcp_time_stamp - tp->rcv_tstamp;
if (tp->keepalive_time > elapsed) if (tp->keepalive_time > elapsed)
elapsed = tp->keepalive_time - elapsed; elapsed = tp->keepalive_time - elapsed;
...@@ -2346,33 +2363,35 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2346,33 +2363,35 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
case TCP_LINGER2: case TCP_LINGER2:
if (val < 0) if (val < 0)
tp->linger2 = -1; tp->linger2 = -1;
else if (val > sysctl_tcp_fin_timeout/HZ) else if (val > sysctl_tcp_fin_timeout / HZ)
tp->linger2 = 0; tp->linger2 = 0;
else else
tp->linger2 = val*HZ; tp->linger2 = val * HZ;
break; break;
case TCP_DEFER_ACCEPT: case TCP_DEFER_ACCEPT:
tp->defer_accept = 0; tp->defer_accept = 0;
if (val > 0) { if (val > 0) {
/* Translate value in seconds to number of retransmits */ /* Translate value in seconds to number of
while (tp->defer_accept < 32 && val > ((TCP_TIMEOUT_INIT/HZ)<<tp->defer_accept)) * retransmits */
while (tp->defer_accept < 32 &&
val > ((TCP_TIMEOUT_INIT / HZ) <<
tp->defer_accept))
tp->defer_accept++; tp->defer_accept++;
tp->defer_accept++; tp->defer_accept++;
} }
break; break;
case TCP_WINDOW_CLAMP: case TCP_WINDOW_CLAMP:
if (val==0) { if (!val) {
if (sk->state != TCP_CLOSE) { if (sk->state != TCP_CLOSE) {
err = -EINVAL; err = -EINVAL;
break; break;
} }
tp->window_clamp = 0; tp->window_clamp = 0;
} else { } else
tp->window_clamp = val<SOCK_MIN_RCVBUF/2 ? tp->window_clamp = val < SOCK_MIN_RCVBUF / 2 ?
SOCK_MIN_RCVBUF/2 : val; SOCK_MIN_RCVBUF / 2 : val;
}
break; break;
case TCP_QUICKACK: case TCP_QUICKACK:
...@@ -2380,7 +2399,8 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2380,7 +2399,8 @@ int tcp_setsockopt(struct sock *sk, int level, int optname, char *optval,
tp->ack.pingpong = 1; tp->ack.pingpong = 1;
} else { } else {
tp->ack.pingpong = 0; tp->ack.pingpong = 0;
if ((1<<sk->state)&(TCPF_ESTABLISHED|TCPF_CLOSE_WAIT) && if ((1 << sk->state) &
(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT) &&
tcp_ack_scheduled(tp)) { tcp_ack_scheduled(tp)) {
tp->ack.pending |= TCP_ACK_PUSHED; tp->ack.pending |= TCP_ACK_PUSHED;
cleanup_rbuf(sk, 1); cleanup_rbuf(sk, 1);
...@@ -2404,22 +2424,22 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2404,22 +2424,22 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
struct tcp_opt *tp = tcp_sk(sk); struct tcp_opt *tp = tcp_sk(sk);
int val, len; int val, len;
if(level != SOL_TCP) if (level != SOL_TCP)
return tp->af_specific->getsockopt(sk, level, optname, return tp->af_specific->getsockopt(sk, level, optname,
optval, optlen); optval, optlen);
if(get_user(len,optlen)) if (get_user(len, optlen))
return -EFAULT; return -EFAULT;
len = min_t(unsigned int, len, sizeof(int)); len = min_t(unsigned int, len, sizeof(int));
if(len < 0) if (len < 0)
return -EINVAL; return -EINVAL;
switch(optname) { switch (optname) {
case TCP_MAXSEG: case TCP_MAXSEG:
val = tp->mss_cache; val = tp->mss_cache;
if (val == 0 && ((1<<sk->state)&(TCPF_CLOSE|TCPF_LISTEN))) if (!val && ((1 << sk->state) & (TCPF_CLOSE | TCPF_LISTEN)))
val = tp->user_mss; val = tp->user_mss;
break; break;
case TCP_NODELAY: case TCP_NODELAY:
...@@ -2429,10 +2449,10 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2429,10 +2449,10 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
val = (tp->nonagle == 2); val = (tp->nonagle == 2);
break; break;
case TCP_KEEPIDLE: case TCP_KEEPIDLE:
val = (tp->keepalive_time ? : sysctl_tcp_keepalive_time)/HZ; val = (tp->keepalive_time ? : sysctl_tcp_keepalive_time) / HZ;
break; break;
case TCP_KEEPINTVL: case TCP_KEEPINTVL:
val = (tp->keepalive_intvl ? : sysctl_tcp_keepalive_intvl)/HZ; val = (tp->keepalive_intvl ? : sysctl_tcp_keepalive_intvl) / HZ;
break; break;
case TCP_KEEPCNT: case TCP_KEEPCNT:
val = tp->keepalive_probes ? : sysctl_tcp_keepalive_probes; val = tp->keepalive_probes ? : sysctl_tcp_keepalive_probes;
...@@ -2443,20 +2463,20 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2443,20 +2463,20 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
case TCP_LINGER2: case TCP_LINGER2:
val = tp->linger2; val = tp->linger2;
if (val >= 0) if (val >= 0)
val = (val ? : sysctl_tcp_fin_timeout)/HZ; val = (val ? : sysctl_tcp_fin_timeout) / HZ;
break; break;
case TCP_DEFER_ACCEPT: case TCP_DEFER_ACCEPT:
val = tp->defer_accept == 0 ? 0 : ((TCP_TIMEOUT_INIT/HZ)<<(tp->defer_accept-1)); val = !tp->defer_accept ? 0 : ((TCP_TIMEOUT_INIT / HZ) <<
(tp->defer_accept - 1));
break; break;
case TCP_WINDOW_CLAMP: case TCP_WINDOW_CLAMP:
val = tp->window_clamp; val = tp->window_clamp;
break; break;
case TCP_INFO: case TCP_INFO: {
{
struct tcp_info info; struct tcp_info info;
u32 now = tcp_time_stamp; u32 now = tcp_time_stamp;
if(get_user(len,optlen)) if (get_user(len, optlen))
return -EFAULT; return -EFAULT;
info.tcpi_state = sk->state; info.tcpi_state = sk->state;
info.tcpi_ca_state = tp->ca_state; info.tcpi_ca_state = tp->ca_state;
...@@ -2476,11 +2496,11 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2476,11 +2496,11 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
info.tcpi_snd_wscale = 0; info.tcpi_snd_wscale = 0;
info.tcpi_rcv_wscale = 0; info.tcpi_rcv_wscale = 0;
} }
if (tp->ecn_flags&TCP_ECN_OK) if (tp->ecn_flags & TCP_ECN_OK)
info.tcpi_options |= TCPI_OPT_ECN; info.tcpi_options |= TCPI_OPT_ECN;
info.tcpi_rto = (1000000*tp->rto)/HZ; info.tcpi_rto = (1000000 * tp->rto) / HZ;
info.tcpi_ato = (1000000*tp->ack.ato)/HZ; info.tcpi_ato = (1000000 * tp->ack.ato) / HZ;
info.tcpi_snd_mss = tp->mss_cache; info.tcpi_snd_mss = tp->mss_cache;
info.tcpi_rcv_mss = tp->ack.rcv_mss; info.tcpi_rcv_mss = tp->ack.rcv_mss;
...@@ -2490,24 +2510,25 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2490,24 +2510,25 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
info.tcpi_retrans = tp->retrans_out; info.tcpi_retrans = tp->retrans_out;
info.tcpi_fackets = tp->fackets_out; info.tcpi_fackets = tp->fackets_out;
info.tcpi_last_data_sent = ((now - tp->lsndtime)*1000)/HZ; info.tcpi_last_data_sent = ((now - tp->lsndtime) * 1000) / HZ;
info.tcpi_last_ack_sent = 0; info.tcpi_last_ack_sent = 0;
info.tcpi_last_data_recv = ((now - tp->ack.lrcvtime)*1000)/HZ; info.tcpi_last_data_recv = ((now -
info.tcpi_last_ack_recv = ((now - tp->rcv_tstamp)*1000)/HZ; tp->ack.lrcvtime) * 1000) / HZ;
info.tcpi_last_ack_recv = ((now - tp->rcv_tstamp) * 1000) / HZ;
info.tcpi_pmtu = tp->pmtu_cookie; info.tcpi_pmtu = tp->pmtu_cookie;
info.tcpi_rcv_ssthresh = tp->rcv_ssthresh; info.tcpi_rcv_ssthresh = tp->rcv_ssthresh;
info.tcpi_rtt = ((1000000*tp->srtt)/HZ)>>3; info.tcpi_rtt = ((1000000 * tp->srtt) / HZ) >> 3;
info.tcpi_rttvar = ((1000000*tp->mdev)/HZ)>>2; info.tcpi_rttvar = ((1000000 * tp->mdev) / HZ) >> 2;
info.tcpi_snd_ssthresh = tp->snd_ssthresh; info.tcpi_snd_ssthresh = tp->snd_ssthresh;
info.tcpi_snd_cwnd = tp->snd_cwnd; info.tcpi_snd_cwnd = tp->snd_cwnd;
info.tcpi_advmss = tp->advmss; info.tcpi_advmss = tp->advmss;
info.tcpi_reordering = tp->reordering; info.tcpi_reordering = tp->reordering;
len = min_t(unsigned int, len, sizeof(info)); len = min_t(unsigned int, len, sizeof(info));
if(put_user(len, optlen)) if (put_user(len, optlen))
return -EFAULT; return -EFAULT;
if(copy_to_user(optval, &info,len)) if (copy_to_user(optval, &info, len))
return -EFAULT; return -EFAULT;
return 0; return 0;
} }
...@@ -2518,9 +2539,9 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval, ...@@ -2518,9 +2539,9 @@ int tcp_getsockopt(struct sock *sk, int level, int optname, char *optval,
return -ENOPROTOOPT; return -ENOPROTOOPT;
}; };
if(put_user(len, optlen)) if (put_user(len, optlen))
return -EFAULT; return -EFAULT;
if(copy_to_user(optval, &val,len)) if (copy_to_user(optval, &val, len))
return -EFAULT; return -EFAULT;
return 0; return 0;
} }
...@@ -2535,7 +2556,7 @@ void __init tcp_init(void) ...@@ -2535,7 +2556,7 @@ void __init tcp_init(void)
unsigned long goal; unsigned long goal;
int order, i; int order, i;
if(sizeof(struct tcp_skb_cb) > sizeof(skb->cb)) if (sizeof(struct tcp_skb_cb) > sizeof(skb->cb))
__skb_cb_too_small_for_tcp(sizeof(struct tcp_skb_cb), __skb_cb_too_small_for_tcp(sizeof(struct tcp_skb_cb),
sizeof(skb->cb)); sizeof(skb->cb));
...@@ -2543,21 +2564,21 @@ void __init tcp_init(void) ...@@ -2543,21 +2564,21 @@ void __init tcp_init(void)
sizeof(struct open_request), sizeof(struct open_request),
0, SLAB_HWCACHE_ALIGN, 0, SLAB_HWCACHE_ALIGN,
NULL, NULL); NULL, NULL);
if(!tcp_openreq_cachep) if (!tcp_openreq_cachep)
panic("tcp_init: Cannot alloc open_request cache."); panic("tcp_init: Cannot alloc open_request cache.");
tcp_bucket_cachep = kmem_cache_create("tcp_bind_bucket", tcp_bucket_cachep = kmem_cache_create("tcp_bind_bucket",
sizeof(struct tcp_bind_bucket), sizeof(struct tcp_bind_bucket),
0, SLAB_HWCACHE_ALIGN, 0, SLAB_HWCACHE_ALIGN,
NULL, NULL); NULL, NULL);
if(!tcp_bucket_cachep) if (!tcp_bucket_cachep)
panic("tcp_init: Cannot alloc tcp_bind_bucket cache."); panic("tcp_init: Cannot alloc tcp_bind_bucket cache.");
tcp_timewait_cachep = kmem_cache_create("tcp_tw_bucket", tcp_timewait_cachep = kmem_cache_create("tcp_tw_bucket",
sizeof(struct tcp_tw_bucket), sizeof(struct tcp_tw_bucket),
0, SLAB_HWCACHE_ALIGN, 0, SLAB_HWCACHE_ALIGN,
NULL, NULL); NULL, NULL);
if(!tcp_timewait_cachep) if (!tcp_timewait_cachep)
panic("tcp_init: Cannot alloc tcp_tw_bucket cache."); panic("tcp_init: Cannot alloc tcp_tw_bucket cache.");
/* Size and allocate the main established and bind bucket /* Size and allocate the main established and bind bucket
...@@ -2570,21 +2591,21 @@ void __init tcp_init(void) ...@@ -2570,21 +2591,21 @@ void __init tcp_init(void)
else else
goal = num_physpages >> (23 - PAGE_SHIFT); goal = num_physpages >> (23 - PAGE_SHIFT);
for(order = 0; (1UL << order) < goal; order++) for (order = 0; (1UL << order) < goal; order++)
; ;
do { do {
tcp_ehash_size = (1UL << order) * PAGE_SIZE / tcp_ehash_size = (1UL << order) * PAGE_SIZE /
sizeof(struct tcp_ehash_bucket); sizeof(struct tcp_ehash_bucket);
tcp_ehash_size >>= 1; tcp_ehash_size >>= 1;
while (tcp_ehash_size & (tcp_ehash_size-1)) while (tcp_ehash_size & (tcp_ehash_size - 1))
tcp_ehash_size--; tcp_ehash_size--;
tcp_ehash = (struct tcp_ehash_bucket *) tcp_ehash = (struct tcp_ehash_bucket *)
__get_free_pages(GFP_ATOMIC, order); __get_free_pages(GFP_ATOMIC, order);
} while (tcp_ehash == NULL && --order > 0); } while (!tcp_ehash && --order > 0);
if (!tcp_ehash) if (!tcp_ehash)
panic("Failed to allocate TCP established hash table\n"); panic("Failed to allocate TCP established hash table\n");
for (i = 0; i < (tcp_ehash_size<<1); i++) { for (i = 0; i < (tcp_ehash_size << 1); i++) {
tcp_ehash[i].lock = RW_LOCK_UNLOCKED; tcp_ehash[i].lock = RW_LOCK_UNLOCKED;
tcp_ehash[i].chain = NULL; tcp_ehash[i].chain = NULL;
} }
...@@ -2596,7 +2617,7 @@ void __init tcp_init(void) ...@@ -2596,7 +2617,7 @@ void __init tcp_init(void)
continue; continue;
tcp_bhash = (struct tcp_bind_hashbucket *) tcp_bhash = (struct tcp_bind_hashbucket *)
__get_free_pages(GFP_ATOMIC, order); __get_free_pages(GFP_ATOMIC, order);
} while (tcp_bhash == NULL && --order >= 0); } while (!tcp_bhash && --order >= 0);
if (!tcp_bhash) if (!tcp_bhash)
panic("Failed to allocate TCP bind hash table\n"); panic("Failed to allocate TCP bind hash table\n");
...@@ -2612,33 +2633,34 @@ void __init tcp_init(void) ...@@ -2612,33 +2633,34 @@ void __init tcp_init(void)
sysctl_local_port_range[0] = 32768; sysctl_local_port_range[0] = 32768;
sysctl_local_port_range[1] = 61000; sysctl_local_port_range[1] = 61000;
sysctl_tcp_max_tw_buckets = 180000; sysctl_tcp_max_tw_buckets = 180000;
sysctl_tcp_max_orphans = 4096<<(order-4); sysctl_tcp_max_orphans = 4096 << (order - 4);
sysctl_max_syn_backlog = 1024; sysctl_max_syn_backlog = 1024;
} else if (order < 3) { } else if (order < 3) {
sysctl_local_port_range[0] = 1024*(3-order); sysctl_local_port_range[0] = 1024 * (3 - order);
sysctl_tcp_max_tw_buckets >>= (3-order); sysctl_tcp_max_tw_buckets >>= (3 - order);
sysctl_tcp_max_orphans >>= (3-order); sysctl_tcp_max_orphans >>= (3 - order);
sysctl_max_syn_backlog = 128; sysctl_max_syn_backlog = 128;
} }
tcp_port_rover = sysctl_local_port_range[0] - 1; tcp_port_rover = sysctl_local_port_range[0] - 1;
sysctl_tcp_mem[0] = 768<<order; sysctl_tcp_mem[0] = 768 << order;
sysctl_tcp_mem[1] = 1024<<order; sysctl_tcp_mem[1] = 1024 << order;
sysctl_tcp_mem[2] = 1536<<order; sysctl_tcp_mem[2] = 1536 << order;
if (sysctl_tcp_mem[2] - sysctl_tcp_mem[1] > 512) if (sysctl_tcp_mem[2] - sysctl_tcp_mem[1] > 512)
sysctl_tcp_mem[1] = sysctl_tcp_mem[2] - 512; sysctl_tcp_mem[1] = sysctl_tcp_mem[2] - 512;
if (sysctl_tcp_mem[1] - sysctl_tcp_mem[0] > 512) if (sysctl_tcp_mem[1] - sysctl_tcp_mem[0] > 512)
sysctl_tcp_mem[0] = sysctl_tcp_mem[1] - 512; sysctl_tcp_mem[0] = sysctl_tcp_mem[1] - 512;
if (order < 3) { if (order < 3) {
sysctl_tcp_wmem[2] = 64*1024; sysctl_tcp_wmem[2] = 64 * 1024;
sysctl_tcp_rmem[0] = PAGE_SIZE; sysctl_tcp_rmem[0] = PAGE_SIZE;
sysctl_tcp_rmem[1] = 43689; sysctl_tcp_rmem[1] = 43689;
sysctl_tcp_rmem[2] = 2*43689; sysctl_tcp_rmem[2] = 2 * 43689;
} }
printk(KERN_INFO "TCP: Hash tables configured (established %d bind %d)\n", printk(KERN_INFO "TCP: Hash tables configured "
tcp_ehash_size<<1, tcp_bhash_size); "(established %d bind %d)\n",
tcp_ehash_size << 1, tcp_bhash_size);
tcpdiag_init(); tcpdiag_init();
} }
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment