Commit 8e33ba49 authored by Linus Torvalds's avatar Linus Torvalds

Merge master.kernel.org:/pub/scm/linux/kernel/git/acme/net-2.6

parents 8cde0776 2d43f112
......@@ -93,6 +93,7 @@ struct tc_fifo_qopt
/* PRIO section */
#define TCQ_PRIO_BANDS 16
#define TCQ_MIN_PRIO_BANDS 2
struct tc_prio_qopt
{
......@@ -169,6 +170,7 @@ struct tc_red_qopt
unsigned char Scell_log; /* cell size for idle damping */
unsigned char flags;
#define TC_RED_ECN 1
#define TC_RED_HARDDROP 2
};
struct tc_red_xstats
......@@ -194,15 +196,11 @@ enum
#define TCA_GRED_MAX (__TCA_GRED_MAX - 1)
#define TCA_SET_OFF TCA_GRED_PARMS
struct tc_gred_qopt
{
__u32 limit; /* HARD maximal queue length (bytes)
*/
__u32 qth_min; /* Min average length threshold (bytes)
*/
__u32 qth_max; /* Max average length threshold (bytes)
*/
__u32 limit; /* HARD maximal queue length (bytes) */
__u32 qth_min; /* Min average length threshold (bytes) */
__u32 qth_max; /* Max average length threshold (bytes) */
__u32 DP; /* upto 2^32 DPs */
__u32 backlog;
__u32 qave;
......@@ -210,22 +208,22 @@ struct tc_gred_qopt
__u32 early;
__u32 other;
__u32 pdrop;
unsigned char Wlog; /* log(W) */
unsigned char Plog; /* log(P_max/(qth_max-qth_min)) */
unsigned char Scell_log; /* cell size for idle damping */
__u8 Wlog; /* log(W) */
__u8 Plog; /* log(P_max/(qth_max-qth_min)) */
__u8 Scell_log; /* cell size for idle damping */
__u8 prio; /* prio of this VQ */
__u32 packets;
__u32 bytesin;
};
/* gred setup */
struct tc_gred_sopt
{
__u32 DPs;
__u32 def_DP;
__u8 grio;
__u8 pad1;
__u16 pad2;
__u8 flags;
__u16 pad1;
};
/* HTB section */
......
......@@ -603,29 +603,46 @@ static inline void skb_queue_head_init(struct sk_buff_head *list)
*/
/**
* __skb_queue_head - queue a buffer at the list head
* __skb_queue_after - queue a buffer at the list head
* @list: list to use
* @prev: place after this buffer
* @newsk: buffer to queue
*
* Queue a buffer at the start of a list. This function takes no locks
* Queue a buffer int the middle of a list. This function takes no locks
* and you must therefore hold required locks before calling it.
*
* A buffer cannot be placed on two lists at the same time.
*/
extern void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
static inline void __skb_queue_head(struct sk_buff_head *list,
static inline void __skb_queue_after(struct sk_buff_head *list,
struct sk_buff *prev,
struct sk_buff *newsk)
{
struct sk_buff *prev, *next;
struct sk_buff *next;
list->qlen++;
prev = (struct sk_buff *)list;
next = prev->next;
newsk->next = next;
newsk->prev = prev;
next->prev = prev->next = newsk;
}
/**
* __skb_queue_head - queue a buffer at the list head
* @list: list to use
* @newsk: buffer to queue
*
* Queue a buffer at the start of a list. This function takes no locks
* and you must therefore hold required locks before calling it.
*
* A buffer cannot be placed on two lists at the same time.
*/
extern void skb_queue_head(struct sk_buff_head *list, struct sk_buff *newsk);
static inline void __skb_queue_head(struct sk_buff_head *list,
struct sk_buff *newsk)
{
__skb_queue_after(list, (struct sk_buff *)list, newsk);
}
/**
* __skb_queue_tail - queue a buffer at the list tail
* @list: list to use
......@@ -1203,6 +1220,11 @@ static inline void kunmap_skb_frag(void *vaddr)
prefetch(skb->next), (skb != (struct sk_buff *)(queue)); \
skb = skb->next)
#define skb_queue_reverse_walk(queue, skb) \
for (skb = (queue)->prev; \
prefetch(skb->prev), (skb != (struct sk_buff *)(queue)); \
skb = skb->prev)
extern struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags,
int noblock, int *err);
......
......@@ -2,6 +2,7 @@
#define _INET_ECN_H_
#include <linux/ip.h>
#include <linux/skbuff.h>
#include <net/dsfield.h>
enum {
......@@ -48,7 +49,7 @@ static inline __u8 INET_ECN_encapsulate(__u8 outer, __u8 inner)
(label) |= __constant_htons(INET_ECN_ECT_0 << 4); \
} while (0)
static inline void IP_ECN_set_ce(struct iphdr *iph)
static inline int IP_ECN_set_ce(struct iphdr *iph)
{
u32 check = iph->check;
u32 ecn = (iph->tos + 1) & INET_ECN_MASK;
......@@ -61,7 +62,7 @@ static inline void IP_ECN_set_ce(struct iphdr *iph)
* INET_ECN_CE => 00
*/
if (!(ecn & 2))
return;
return !ecn;
/*
* The following gives us:
......@@ -72,6 +73,7 @@ static inline void IP_ECN_set_ce(struct iphdr *iph)
iph->check = check + (check>=0xFFFF);
iph->tos |= INET_ECN_CE;
return 1;
}
static inline void IP_ECN_clear(struct iphdr *iph)
......@@ -87,11 +89,12 @@ static inline void ipv4_copy_dscp(struct iphdr *outer, struct iphdr *inner)
struct ipv6hdr;
static inline void IP6_ECN_set_ce(struct ipv6hdr *iph)
static inline int IP6_ECN_set_ce(struct ipv6hdr *iph)
{
if (INET_ECN_is_not_ect(ipv6_get_dsfield(iph)))
return;
return 0;
*(u32*)iph |= htonl(INET_ECN_CE << 20);
return 1;
}
static inline void IP6_ECN_clear(struct ipv6hdr *iph)
......@@ -105,4 +108,21 @@ static inline void ipv6_copy_dscp(struct ipv6hdr *outer, struct ipv6hdr *inner)
ipv6_change_dsfield(inner, INET_ECN_MASK, dscp);
}
static inline int INET_ECN_set_ce(struct sk_buff *skb)
{
switch (skb->protocol) {
case __constant_htons(ETH_P_IP):
if (skb->nh.raw + sizeof(struct iphdr) <= skb->tail)
return IP_ECN_set_ce(skb->nh.iph);
break;
case __constant_htons(ETH_P_IPV6):
if (skb->nh.raw + sizeof(struct ipv6hdr) <= skb->tail)
return IP6_ECN_set_ce(skb->nh.ipv6h);
break;
}
return 0;
}
#endif
......@@ -125,9 +125,7 @@ struct inet_hashinfo {
rwlock_t lhash_lock ____cacheline_aligned;
atomic_t lhash_users;
wait_queue_head_t lhash_wait;
spinlock_t portalloc_lock;
kmem_cache_t *bind_bucket_cachep;
int port_rover;
};
static inline unsigned int inet_ehashfn(const __u32 laddr, const __u16 lport,
......
#ifndef __NET_SCHED_RED_H
#define __NET_SCHED_RED_H
#include <linux/config.h>
#include <linux/types.h>
#include <net/pkt_sched.h>
#include <net/inet_ecn.h>
#include <net/dsfield.h>
/* Random Early Detection (RED) algorithm.
=======================================
Source: Sally Floyd and Van Jacobson, "Random Early Detection Gateways
for Congestion Avoidance", 1993, IEEE/ACM Transactions on Networking.
This file codes a "divisionless" version of RED algorithm
as written down in Fig.17 of the paper.
Short description.
------------------
When a new packet arrives we calculate the average queue length:
avg = (1-W)*avg + W*current_queue_len,
W is the filter time constant (chosen as 2^(-Wlog)), it controls
the inertia of the algorithm. To allow larger bursts, W should be
decreased.
if (avg > th_max) -> packet marked (dropped).
if (avg < th_min) -> packet passes.
if (th_min < avg < th_max) we calculate probability:
Pb = max_P * (avg - th_min)/(th_max-th_min)
and mark (drop) packet with this probability.
Pb changes from 0 (at avg==th_min) to max_P (avg==th_max).
max_P should be small (not 1), usually 0.01..0.02 is good value.
max_P is chosen as a number, so that max_P/(th_max-th_min)
is a negative power of two in order arithmetics to contain
only shifts.
Parameters, settable by user:
-----------------------------
qth_min - bytes (should be < qth_max/2)
qth_max - bytes (should be at least 2*qth_min and less limit)
Wlog - bits (<32) log(1/W).
Plog - bits (<32)
Plog is related to max_P by formula:
max_P = (qth_max-qth_min)/2^Plog;
F.e. if qth_max=128K and qth_min=32K, then Plog=22
corresponds to max_P=0.02
Scell_log
Stab
Lookup table for log((1-W)^(t/t_ave).
NOTES:
Upper bound on W.
-----------------
If you want to allow bursts of L packets of size S,
you should choose W:
L + 1 - th_min/S < (1-(1-W)^L)/W
th_min/S = 32 th_min/S = 4
log(W) L
-1 33
-2 35
-3 39
-4 46
-5 57
-6 75
-7 101
-8 135
-9 190
etc.
*/
#define RED_STAB_SIZE 256
#define RED_STAB_MASK (RED_STAB_SIZE - 1)
struct red_stats
{
u32 prob_drop; /* Early probability drops */
u32 prob_mark; /* Early probability marks */
u32 forced_drop; /* Forced drops, qavg > max_thresh */
u32 forced_mark; /* Forced marks, qavg > max_thresh */
u32 pdrop; /* Drops due to queue limits */
u32 other; /* Drops due to drop() calls */
u32 backlog;
};
struct red_parms
{
/* Parameters */
u32 qth_min; /* Min avg length threshold: A scaled */
u32 qth_max; /* Max avg length threshold: A scaled */
u32 Scell_max;
u32 Rmask; /* Cached random mask, see red_rmask */
u8 Scell_log;
u8 Wlog; /* log(W) */
u8 Plog; /* random number bits */
u8 Stab[RED_STAB_SIZE];
/* Variables */
int qcount; /* Number of packets since last random
number generation */
u32 qR; /* Cached random number */
unsigned long qavg; /* Average queue length: A scaled */
psched_time_t qidlestart; /* Start of current idle period */
};
static inline u32 red_rmask(u8 Plog)
{
return Plog < 32 ? ((1 << Plog) - 1) : ~0UL;
}
static inline void red_set_parms(struct red_parms *p,
u32 qth_min, u32 qth_max, u8 Wlog, u8 Plog,
u8 Scell_log, u8 *stab)
{
/* Reset average queue length, the value is strictly bound
* to the parameters below, reseting hurts a bit but leaving
* it might result in an unreasonable qavg for a while. --TGR
*/
p->qavg = 0;
p->qcount = -1;
p->qth_min = qth_min << Wlog;
p->qth_max = qth_max << Wlog;
p->Wlog = Wlog;
p->Plog = Plog;
p->Rmask = red_rmask(Plog);
p->Scell_log = Scell_log;
p->Scell_max = (255 << Scell_log);
memcpy(p->Stab, stab, sizeof(p->Stab));
}
static inline int red_is_idling(struct red_parms *p)
{
return !PSCHED_IS_PASTPERFECT(p->qidlestart);
}
static inline void red_start_of_idle_period(struct red_parms *p)
{
PSCHED_GET_TIME(p->qidlestart);
}
static inline void red_end_of_idle_period(struct red_parms *p)
{
PSCHED_SET_PASTPERFECT(p->qidlestart);
}
static inline void red_restart(struct red_parms *p)
{
red_end_of_idle_period(p);
p->qavg = 0;
p->qcount = -1;
}
static inline unsigned long red_calc_qavg_from_idle_time(struct red_parms *p)
{
psched_time_t now;
long us_idle;
int shift;
PSCHED_GET_TIME(now);
us_idle = PSCHED_TDIFF_SAFE(now, p->qidlestart, p->Scell_max);
/*
* The problem: ideally, average length queue recalcultion should
* be done over constant clock intervals. This is too expensive, so
* that the calculation is driven by outgoing packets.
* When the queue is idle we have to model this clock by hand.
*
* SF+VJ proposed to "generate":
*
* m = idletime / (average_pkt_size / bandwidth)
*
* dummy packets as a burst after idle time, i.e.
*
* p->qavg *= (1-W)^m
*
* This is an apparently overcomplicated solution (f.e. we have to
* precompute a table to make this calculation in reasonable time)
* I believe that a simpler model may be used here,
* but it is field for experiments.
*/
shift = p->Stab[(us_idle >> p->Scell_log) & RED_STAB_MASK];
if (shift)
return p->qavg >> shift;
else {
/* Approximate initial part of exponent with linear function:
*
* (1-W)^m ~= 1-mW + ...
*
* Seems, it is the best solution to
* problem of too coarse exponent tabulation.
*/
us_idle = (p->qavg * us_idle) >> p->Scell_log;
if (us_idle < (p->qavg >> 1))
return p->qavg - us_idle;
else
return p->qavg >> 1;
}
}
static inline unsigned long red_calc_qavg_no_idle_time(struct red_parms *p,
unsigned int backlog)
{
/*
* NOTE: p->qavg is fixed point number with point at Wlog.
* The formula below is equvalent to floating point
* version:
*
* qavg = qavg*(1-W) + backlog*W;
*
* --ANK (980924)
*/
return p->qavg + (backlog - (p->qavg >> p->Wlog));
}
static inline unsigned long red_calc_qavg(struct red_parms *p,
unsigned int backlog)
{
if (!red_is_idling(p))
return red_calc_qavg_no_idle_time(p, backlog);
else
return red_calc_qavg_from_idle_time(p);
}
static inline u32 red_random(struct red_parms *p)
{
return net_random() & p->Rmask;
}
static inline int red_mark_probability(struct red_parms *p, unsigned long qavg)
{
/* The formula used below causes questions.
OK. qR is random number in the interval 0..Rmask
i.e. 0..(2^Plog). If we used floating point
arithmetics, it would be: (2^Plog)*rnd_num,
where rnd_num is less 1.
Taking into account, that qavg have fixed
point at Wlog, and Plog is related to max_P by
max_P = (qth_max-qth_min)/2^Plog; two lines
below have the following floating point equivalent:
max_P*(qavg - qth_min)/(qth_max-qth_min) < rnd/qcount
Any questions? --ANK (980924)
*/
return !(((qavg - p->qth_min) >> p->Wlog) * p->qcount < p->qR);
}
enum {
RED_BELOW_MIN_THRESH,
RED_BETWEEN_TRESH,
RED_ABOVE_MAX_TRESH,
};
static inline int red_cmp_thresh(struct red_parms *p, unsigned long qavg)
{
if (qavg < p->qth_min)
return RED_BELOW_MIN_THRESH;
else if (qavg >= p->qth_max)
return RED_ABOVE_MAX_TRESH;
else
return RED_BETWEEN_TRESH;
}
enum {
RED_DONT_MARK,
RED_PROB_MARK,
RED_HARD_MARK,
};
static inline int red_action(struct red_parms *p, unsigned long qavg)
{
switch (red_cmp_thresh(p, qavg)) {
case RED_BELOW_MIN_THRESH:
p->qcount = -1;
return RED_DONT_MARK;
case RED_BETWEEN_TRESH:
if (++p->qcount) {
if (red_mark_probability(p, qavg)) {
p->qcount = 0;
p->qR = red_random(p);
return RED_PROB_MARK;
}
} else
p->qR = red_random(p);
return RED_DONT_MARK;
case RED_ABOVE_MAX_TRESH:
p->qcount = -1;
return RED_HARD_MARK;
}
BUG();
return RED_DONT_MARK;
}
#endif
......@@ -52,8 +52,9 @@ int sk_stream_wait_connect(struct sock *sk, long *timeo_p)
{
struct task_struct *tsk = current;
DEFINE_WAIT(wait);
int done;
while (1) {
do {
if (sk->sk_err)
return sock_error(sk);
if ((1 << sk->sk_state) & ~(TCPF_SYN_SENT | TCPF_SYN_RECV))
......@@ -65,13 +66,12 @@ int sk_stream_wait_connect(struct sock *sk, long *timeo_p)
prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
sk->sk_write_pending++;
if (sk_wait_event(sk, timeo_p,
done = sk_wait_event(sk, timeo_p,
!((1 << sk->sk_state) &
~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT))))
break;
~(TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)));
finish_wait(sk->sk_sleep, &wait);
sk->sk_write_pending--;
}
} while (!done);
return 0;
}
......
......@@ -31,8 +31,6 @@ struct inet_hashinfo __cacheline_aligned dccp_hashinfo = {
.lhash_lock = RW_LOCK_UNLOCKED,
.lhash_users = ATOMIC_INIT(0),
.lhash_wait = __WAIT_QUEUE_HEAD_INITIALIZER(dccp_hashinfo.lhash_wait),
.portalloc_lock = SPIN_LOCK_UNLOCKED,
.port_rover = 1024 - 1,
};
EXPORT_SYMBOL_GPL(dccp_hashinfo);
......@@ -125,36 +123,15 @@ static int dccp_v4_hash_connect(struct sock *sk)
int ret;
if (snum == 0) {
int rover;
int low = sysctl_local_port_range[0];
int high = sysctl_local_port_range[1];
int remaining = (high - low) + 1;
int rover = net_random() % (high - low) + low;
struct hlist_node *node;
struct inet_timewait_sock *tw = NULL;
local_bh_disable();
/* TODO. Actually it is not so bad idea to remove
* dccp_hashinfo.portalloc_lock before next submission to
* Linus.
* As soon as we touch this place at all it is time to think.
*
* Now it protects single _advisory_ variable
* dccp_hashinfo.port_rover, hence it is mostly useless.
* Code will work nicely if we just delete it, but
* I am afraid in contented case it will work not better or
* even worse: another cpu just will hit the same bucket
* and spin there.
* So some cpu salt could remove both contention and
* memory pingpong. Any ideas how to do this in a nice way?
*/
spin_lock(&dccp_hashinfo.portalloc_lock);
rover = dccp_hashinfo.port_rover;
do {
rover++;
if ((rover < low) || (rover > high))
rover = low;
head = &dccp_hashinfo.bhash[inet_bhashfn(rover,
dccp_hashinfo.bhash_size)];
spin_lock(&head->lock);
......@@ -187,9 +164,9 @@ static int dccp_v4_hash_connect(struct sock *sk)
next_port:
spin_unlock(&head->lock);
if (++rover > high)
rover = low;
} while (--remaining > 0);
dccp_hashinfo.port_rover = rover;
spin_unlock(&dccp_hashinfo.portalloc_lock);
local_bh_enable();
......@@ -197,9 +174,6 @@ static int dccp_v4_hash_connect(struct sock *sk)
ok:
/* All locks still held and bhs disabled */
dccp_hashinfo.port_rover = rover;
spin_unlock(&dccp_hashinfo.portalloc_lock);
inet_bind_hash(sk, tb, rover);
if (sk_unhashed(sk)) {
inet_sk(sk)->sport = htons(rover);
......
......@@ -78,17 +78,9 @@ int inet_csk_get_port(struct inet_hashinfo *hashinfo,
int low = sysctl_local_port_range[0];
int high = sysctl_local_port_range[1];
int remaining = (high - low) + 1;
int rover;
int rover = net_random() % (high - low) + low;
spin_lock(&hashinfo->portalloc_lock);
if (hashinfo->port_rover < low)
rover = low;
else
rover = hashinfo->port_rover;
do {
rover++;
if (rover > high)
rover = low;
head = &hashinfo->bhash[inet_bhashfn(rover, hashinfo->bhash_size)];
spin_lock(&head->lock);
inet_bind_bucket_for_each(tb, node, &head->chain)
......@@ -97,9 +89,9 @@ int inet_csk_get_port(struct inet_hashinfo *hashinfo,
break;
next:
spin_unlock(&head->lock);
if (++rover > high)
rover = low;
} while (--remaining > 0);
hashinfo->port_rover = rover;
spin_unlock(&hashinfo->portalloc_lock);
/* Exhausted local port range during search? It is not
* possible for us to be holding one of the bind hash
......
......@@ -270,14 +270,10 @@ exp_gre(struct ip_conntrack *master,
exp_orig->expectfn = pptp_expectfn;
exp_orig->flags = 0;
exp_orig->dir = IP_CT_DIR_ORIGINAL;
/* both expectations are identical apart from tuple */
memcpy(exp_reply, exp_orig, sizeof(*exp_reply));
memcpy(&exp_reply->tuple, &exp_tuples[1], sizeof(exp_reply->tuple));
exp_reply->dir = !exp_orig->dir;
if (ip_nat_pptp_hook_exp_gre)
ret = ip_nat_pptp_hook_exp_gre(exp_orig, exp_reply);
else {
......
......@@ -815,7 +815,7 @@ ctnetlink_get_conntrack(struct sock *ctnl, struct sk_buff *skb,
IPCTNL_MSG_CT_NEW, 1, ct);
ip_conntrack_put(ct);
if (err <= 0)
goto out;
goto free;
err = netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
if (err < 0)
......@@ -824,9 +824,9 @@ ctnetlink_get_conntrack(struct sock *ctnl, struct sk_buff *skb,
DEBUGP("leaving\n");
return 0;
out:
if (skb2)
free:
kfree_skb(skb2);
out:
return -1;
}
......@@ -1322,21 +1322,16 @@ ctnetlink_get_expect(struct sock *ctnl, struct sk_buff *skb,
nlh->nlmsg_seq, IPCTNL_MSG_EXP_NEW,
1, exp);
if (err <= 0)
goto out;
goto free;
ip_conntrack_expect_put(exp);
err = netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
if (err < 0)
goto free;
return err;
return netlink_unicast(ctnl, skb2, NETLINK_CB(skb).pid, MSG_DONTWAIT);
out:
ip_conntrack_expect_put(exp);
free:
if (skb2)
kfree_skb(skb2);
out:
ip_conntrack_expect_put(exp);
return err;
}
......
......@@ -66,10 +66,8 @@ ip_nat_proto_find_get(u_int8_t protonum)
* removed until we've grabbed the reference */
preempt_disable();
p = __ip_nat_proto_find(protonum);
if (p) {
if (!try_module_get(p->me))
p = &ip_nat_unknown_protocol;
}
preempt_enable();
return p;
......
......@@ -216,6 +216,7 @@ pptp_exp_gre(struct ip_conntrack_expect *expect_orig,
expect_orig->saved_proto.gre.key = htons(nat_pptp_info->pac_call_id);
expect_orig->tuple.src.u.gre.key = htons(nat_pptp_info->pns_call_id);
expect_orig->tuple.dst.u.gre.key = htons(ct_pptp_info->pac_call_id);
expect_orig->dir = IP_CT_DIR_ORIGINAL;
inv_t.src.ip = reply_t->src.ip;
inv_t.dst.ip = reply_t->dst.ip;
inv_t.src.u.gre.key = htons(nat_pptp_info->pac_call_id);
......@@ -233,6 +234,7 @@ pptp_exp_gre(struct ip_conntrack_expect *expect_orig,
expect_reply->saved_proto.gre.key = htons(nat_pptp_info->pns_call_id);
expect_reply->tuple.src.u.gre.key = htons(nat_pptp_info->pac_call_id);
expect_reply->tuple.dst.u.gre.key = htons(ct_pptp_info->pns_call_id);
expect_reply->dir = IP_CT_DIR_REPLY;
inv_t.src.ip = orig_t->src.ip;
inv_t.dst.ip = orig_t->dst.ip;
inv_t.src.u.gre.key = htons(nat_pptp_info->pns_call_id);
......
......@@ -139,8 +139,8 @@ gre_manip_pkt(struct sk_buff **pskb,
break;
case GRE_VERSION_PPTP:
DEBUGP("call_id -> 0x%04x\n",
ntohl(tuple->dst.u.gre.key));
pgreh->call_id = htons(ntohl(tuple->dst.u.gre.key));
ntohs(tuple->dst.u.gre.key));
pgreh->call_id = tuple->dst.u.gre.key;
break;
default:
DEBUGP("can't nat unknown GRE version\n");
......
......@@ -62,7 +62,7 @@ unknown_print_range(char *buffer, const struct ip_nat_range *range)
struct ip_nat_protocol ip_nat_unknown_protocol = {
.name = "unknown",
.me = THIS_MODULE,
/* .me isn't set: getting a ref to this cannot fail. */
.manip_pkt = unknown_manip_pkt,
.in_range = unknown_in_range,
.unique_tuple = unknown_unique_tuple,
......
......@@ -109,6 +109,7 @@ static struct ipt_target ipt_connmark_reg = {
static int __init init(void)
{
need_ip_conntrack();
return ipt_register_target(&ipt_connmark_reg);
}
......
......@@ -2112,7 +2112,6 @@ void __init tcp_init(void)
sysctl_tcp_max_orphans >>= (3 - order);
sysctl_max_syn_backlog = 128;
}
tcp_hashinfo.port_rover = sysctl_local_port_range[0] - 1;
sysctl_tcp_mem[0] = 768 << order;
sysctl_tcp_mem[1] = 1024 << order;
......
......@@ -93,8 +93,6 @@ struct inet_hashinfo __cacheline_aligned tcp_hashinfo = {
.lhash_lock = RW_LOCK_UNLOCKED,
.lhash_users = ATOMIC_INIT(0),
.lhash_wait = __WAIT_QUEUE_HEAD_INITIALIZER(tcp_hashinfo.lhash_wait),
.portalloc_lock = SPIN_LOCK_UNLOCKED,
.port_rover = 1024 - 1,
};
static int tcp_v4_get_port(struct sock *sk, unsigned short snum)
......
......@@ -114,16 +114,9 @@ static int tcp_v6_get_port(struct sock *sk, unsigned short snum)
int low = sysctl_local_port_range[0];
int high = sysctl_local_port_range[1];
int remaining = (high - low) + 1;
int rover;
int rover = net_random() % (high - low) + low;
spin_lock(&tcp_hashinfo.portalloc_lock);
if (tcp_hashinfo.port_rover < low)
rover = low;
else
rover = tcp_hashinfo.port_rover;
do { rover++;
if (rover > high)
rover = low;
do {
head = &tcp_hashinfo.bhash[inet_bhashfn(rover, tcp_hashinfo.bhash_size)];
spin_lock(&head->lock);
inet_bind_bucket_for_each(tb, node, &head->chain)
......@@ -132,9 +125,9 @@ static int tcp_v6_get_port(struct sock *sk, unsigned short snum)
break;
next:
spin_unlock(&head->lock);
if (++rover > high)
rover = low;
} while (--remaining > 0);
tcp_hashinfo.port_rover = rover;
spin_unlock(&tcp_hashinfo.portalloc_lock);
/* Exhausted local port range during search? It is not
* possible for us to be holding one of the bind hash
......
......@@ -117,7 +117,7 @@ int nf_queue(struct sk_buff **skb,
/* QUEUE == DROP if noone is waiting, to be safe. */
read_lock(&queue_handler_lock);
if (!queue_handler[pf]->outfn) {
if (!queue_handler[pf] || !queue_handler[pf]->outfn) {
read_unlock(&queue_handler_lock);
kfree_skb(*skb);
return 1;
......
......@@ -146,11 +146,10 @@ instance_create(u_int16_t group_num, int pid)
goto out_unlock;
}
inst = kmalloc(sizeof(*inst), GFP_ATOMIC);
inst = kzalloc(sizeof(*inst), GFP_ATOMIC);
if (!inst)
goto out_unlock;
memset(inst, 0, sizeof(*inst));
INIT_HLIST_NODE(&inst->hlist);
inst->lock = SPIN_LOCK_UNLOCKED;
/* needs to be two, since we _put() after creation */
......@@ -962,10 +961,9 @@ static int nful_open(struct inode *inode, struct file *file)
struct iter_state *is;
int ret;
is = kmalloc(sizeof(*is), GFP_KERNEL);
is = kzalloc(sizeof(*is), GFP_KERNEL);
if (!is)
return -ENOMEM;
memset(is, 0, sizeof(*is));
ret = seq_open(file, &nful_seq_ops);
if (ret < 0)
goto out_free;
......
......@@ -136,11 +136,10 @@ instance_create(u_int16_t queue_num, int pid)
goto out_unlock;
}
inst = kmalloc(sizeof(*inst), GFP_ATOMIC);
inst = kzalloc(sizeof(*inst), GFP_ATOMIC);
if (!inst)
goto out_unlock;
memset(inst, 0, sizeof(*inst));
inst->queue_num = queue_num;
inst->peer_pid = pid;
inst->queue_maxlen = NFQNL_QMAX_DEFAULT;
......@@ -1036,10 +1035,9 @@ static int nfqnl_open(struct inode *inode, struct file *file)
struct iter_state *is;
int ret;
is = kmalloc(sizeof(*is), GFP_KERNEL);
is = kzalloc(sizeof(*is), GFP_KERNEL);
if (!is)
return -ENOMEM;
memset(is, 0, sizeof(*is));
ret = seq_open(file, &nfqnl_seq_ops);
if (ret < 0)
goto out_free;
......
This diff is collapsed.
......@@ -25,6 +25,8 @@
#include <net/pkt_sched.h>
#define VERSION "1.1"
/* Network Emulation Queuing algorithm.
====================================
......@@ -185,10 +187,13 @@ static int netem_enqueue(struct sk_buff *skb, struct Qdisc *sch)
|| q->counter < q->gap /* inside last reordering gap */
|| q->reorder < get_crandom(&q->reorder_cor)) {
psched_time_t now;
psched_tdiff_t delay;
delay = tabledist(q->latency, q->jitter,
&q->delay_cor, q->delay_dist);
PSCHED_GET_TIME(now);
PSCHED_TADD2(now, tabledist(q->latency, q->jitter,
&q->delay_cor, q->delay_dist),
cb->time_to_send);
PSCHED_TADD2(now, delay, cb->time_to_send);
++q->counter;
ret = q->qdisc->enqueue(skb, q->qdisc);
} else {
......@@ -248,24 +253,31 @@ static struct sk_buff *netem_dequeue(struct Qdisc *sch)
const struct netem_skb_cb *cb
= (const struct netem_skb_cb *)skb->cb;
psched_time_t now;
long delay;
/* if more time remaining? */
PSCHED_GET_TIME(now);
delay = PSCHED_US2JIFFIE(PSCHED_TDIFF(cb->time_to_send, now));
pr_debug("netem_run: skb=%p delay=%ld\n", skb, delay);
if (delay <= 0) {
if (PSCHED_TLESS(cb->time_to_send, now)) {
pr_debug("netem_dequeue: return skb=%p\n", skb);
sch->q.qlen--;
sch->flags &= ~TCQ_F_THROTTLED;
return skb;
} else {
psched_tdiff_t delay = PSCHED_TDIFF(cb->time_to_send, now);
if (q->qdisc->ops->requeue(skb, q->qdisc) != NET_XMIT_SUCCESS) {
sch->qstats.drops++;
/* After this qlen is confused */
printk(KERN_ERR "netem: queue discpline %s could not requeue\n",
q->qdisc->ops->id);
sch->q.qlen--;
}
mod_timer(&q->timer, jiffies + delay);
mod_timer(&q->timer, jiffies + PSCHED_US2JIFFIE(delay));
sch->flags |= TCQ_F_THROTTLED;
if (q->qdisc->ops->requeue(skb, q->qdisc) != 0)
sch->qstats.drops++;
}
}
return NULL;
......@@ -290,11 +302,16 @@ static void netem_reset(struct Qdisc *sch)
del_timer_sync(&q->timer);
}
/* Pass size change message down to embedded FIFO */
static int set_fifo_limit(struct Qdisc *q, int limit)
{
struct rtattr *rta;
int ret = -ENOMEM;
/* Hack to avoid sending change message to non-FIFO */
if (strncmp(q->ops->id + 1, "fifo", 4) != 0)
return 0;
rta = kmalloc(RTA_LENGTH(sizeof(struct tc_fifo_qopt)), GFP_KERNEL);
if (rta) {
rta->rta_type = RTM_NEWQDISC;
......@@ -426,6 +443,84 @@ static int netem_change(struct Qdisc *sch, struct rtattr *opt)
return 0;
}
/*
* Special case version of FIFO queue for use by netem.
* It queues in order based on timestamps in skb's
*/
struct fifo_sched_data {
u32 limit;
};
static int tfifo_enqueue(struct sk_buff *nskb, struct Qdisc *sch)
{
struct fifo_sched_data *q = qdisc_priv(sch);
struct sk_buff_head *list = &sch->q;
const struct netem_skb_cb *ncb
= (const struct netem_skb_cb *)nskb->cb;
struct sk_buff *skb;
if (likely(skb_queue_len(list) < q->limit)) {
skb_queue_reverse_walk(list, skb) {
const struct netem_skb_cb *cb
= (const struct netem_skb_cb *)skb->cb;
if (PSCHED_TLESS(cb->time_to_send, ncb->time_to_send))
break;
}
__skb_queue_after(list, skb, nskb);
sch->qstats.backlog += nskb->len;
sch->bstats.bytes += nskb->len;
sch->bstats.packets++;
return NET_XMIT_SUCCESS;
}
return qdisc_drop(nskb, sch);
}
static int tfifo_init(struct Qdisc *sch, struct rtattr *opt)
{
struct fifo_sched_data *q = qdisc_priv(sch);
if (opt) {
struct tc_fifo_qopt *ctl = RTA_DATA(opt);
if (RTA_PAYLOAD(opt) < sizeof(*ctl))
return -EINVAL;
q->limit = ctl->limit;
} else
q->limit = max_t(u32, sch->dev->tx_queue_len, 1);
return 0;
}
static int tfifo_dump(struct Qdisc *sch, struct sk_buff *skb)
{
struct fifo_sched_data *q = qdisc_priv(sch);
struct tc_fifo_qopt opt = { .limit = q->limit };
RTA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt);
return skb->len;
rtattr_failure:
return -1;
}
static struct Qdisc_ops tfifo_qdisc_ops = {
.id = "tfifo",
.priv_size = sizeof(struct fifo_sched_data),
.enqueue = tfifo_enqueue,
.dequeue = qdisc_dequeue_head,
.requeue = qdisc_requeue,
.drop = qdisc_queue_drop,
.init = tfifo_init,
.reset = qdisc_reset_queue,
.change = tfifo_init,
.dump = tfifo_dump,
};
static int netem_init(struct Qdisc *sch, struct rtattr *opt)
{
struct netem_sched_data *q = qdisc_priv(sch);
......@@ -438,7 +533,7 @@ static int netem_init(struct Qdisc *sch, struct rtattr *opt)
q->timer.function = netem_watchdog;
q->timer.data = (unsigned long) sch;
q->qdisc = qdisc_create_dflt(sch->dev, &pfifo_qdisc_ops);
q->qdisc = qdisc_create_dflt(sch->dev, &tfifo_qdisc_ops);
if (!q->qdisc) {
pr_debug("netem: qdisc create failed\n");
return -ENOMEM;
......@@ -601,6 +696,7 @@ static struct Qdisc_ops netem_qdisc_ops = {
static int __init netem_module_init(void)
{
pr_info("netem: version " VERSION "\n");
return register_qdisc(&netem_qdisc_ops);
}
static void __exit netem_module_exit(void)
......
This diff is collapsed.
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